The Persian Language in Yuan-Dynasty China: A Reappraisal

Stephen G. Haw

It has often been claimed that Persian was an important lingua franca in the Yuan empire. A recent article by Professor David Morgan has discussed this premise at some length, setting out what seems to be impressive evidence in its favour.[1] For some time, however, I have entertained doubts about the validity of some of this evidence. Although I have no doubt that there were a significant number of Persian speakers in the Yuan empire, of whom a number may have held important official positions, I believe that the Persian language was never a genuine lingua franca in China and Mongolia. Its use was probably confined to a section of the Muslim community, and to limited commercial and official circles. Its precise importance must have varied over time, but, generally speaking, other languages were of higher status and more commonly used. Mongolian, the language of the rulers, undoubtedly held the highest status. Turkic was almost certainly in more common usage than Persian, not only in the Yuan empire, but throughout most of the Yeke Mongghol Ulus. It must be realised that most of the variants of Turkic, whether Uighur, Cuman (Qipchaq), Qangli, or whatever, were mutually intelligible.[2] Thus, from the Uighur lands around the Tarim Basin all the way to the Black Sea, what was effectively a single language was predominant.[3]

First, I shall examine the evidence adduced for the importance of Persian. Professor Morgan begins his article with the common claim that Marco Polo must have had a knowledge of Persian, which was ‘spoken and written very widely in China (and elsewhere in the Mongol empire), in the circles in which Marco Polo moved’.[4] Morgan, however, is a Persian scholar, with no knowledge of Chinese. As he has no direct access to Chinese sources in the original language,[5] it must be assumed that he is here following the opinions of others. He is undoubtedly in very good company. The idea that Marco Polo was only really fluent in Persian, and certainly had no knowledge of Chinese, can be traced back at least to the 1870s. In his heavily annotated edition of Marco Polo’s book, first published in 1871, Colonel Sir Henry Yule states that there are ‘positive indications of Marco’s ignorance of Chinese’,[6] and that ‘his intercourse and conversation … at the Kaan’s court … probably was carried on in the Persian language’.[7] One wonders how Yule could have come to such a conclusion, however. He himself certainly was ignorant of the Chinese language, and apparently knew little about China, to judge by his heavy reliance on eighteenth- and nineteenth-century Western accounts of the country.[8] The great Mongolist and Chinese scholar Francis Cleaves commented: ‘I think that Yule was somewhat harsh in his appraisal of Marco Polo’s knowledge of Chinese …’.[9] It may be noted also that Pauthier, who did have knowledge of Chinese, considered that Marco may well have learned several languages of the Yuan court, including both Persian and Chinese, and that he probably also learned the Chinese, Uighur, 'Phags-pa and Perso-Arabic scripts.[10] All this is highly speculative, of course.

Great weight was added to the theory that Marco Polo principally, if not entirely, relied on a knowledge of Persian while in the Yuan empire, by no less an authority than Paul Pelliot. He refers approvingly to Yule’s opinion, and avers that: ‘Marco Polo, maint exemple le prouve, était entouré de Persans … : le persan est même sans doute la seule langue orientale qu’il ait jamais maniée couramment’.[11] This opinion was taken up by Henri Cordier, who quotes Pelliot in his Notes and Addenda to Yule’s edition of Marco Polo’s book.[12] It is also repeated in Pelliot’s Notes on Marco Polo, where Pelliot several times says that Marco ‘had in mind, as usual, the Persian term’,[13] or something similar. Pelliot’s opinion has naturally carried a great deal of weight. Frances Wood took up the refrain, listing, as Persianised versions of Chinese place-names, Polo’s ‘Chemeinfu for Kaipingfu, Pianfu for Pingyangfu, Quengianfu for Xi’anfu and Taianfu for Taiyuanfu’.[14] This is a very short list, but at least a couple more names could be added to it. A couple could also be removed from it, however, as will be shown shortly. In reality, Marco Polo uses only a very few clearly Persian versions of Chinese place-names, and other Persian vocabulary. Most of the names that he gives for places in China are purely Chinese:[15] the fact that some of them are used in more or less the same forms by Rashīd al-Dīn and other Persian writers does not make them any less so. In some cases, too, he uses what are very clearly Mongolian and Turkic versions of place-names, and a number of words from various languages other than Persian.[16]

Wood’s list requires close examination. ‘Chemeinfu’ is undoubtedly a Persian version of Kaipingfu 開平府. Marco uses this name for the city only once, however, when it is first mentioned in his book. This is when he is recounting how he first arrived at the court of Qubilai Qa’an, with his father and uncle.[17] This may explain the use of the Persian version of the name at this point: Marco had only just arrived in the Yuan empire and, at the time, probably only knew Persian, which he had learned during the outward journey. He had not had time to learn any of the languages of Yuan China. If he really did make notes during his travels, and used them when recalling his adventures for Rustichello to record, then he may well have stuck to names as he had noted them at the time.[18] Apart from this one usage, he always calls this city ‘Ciandu’ (or something similar, as there are, as usual, a number of scribal variants in different manuscripts of the book). This was derived from the Chinese ‘Shangdu’ 上都, and is not in any way Persian. Even Pelliot had to admit that he could ‘not find the name in Rašīdu-‘d-Dīn, who uses only Keminfu’.[19] This, then, is a clear example of Marco not following Persian practice. Incidentally, Pelliot is wrong to say that ‘the name of K’ai-p’ing-fu was changed to Shang-tu’ in 1263. The superior prefecture (fu 府), and also the local county (xian 縣), continued to be called Kaiping. Shangdu was an additional name, which recognised the status of the city as an imperial residence. It was used for the route (lu 路) of which the city was the centre of government, but Kaipingfu was not immediately abolished.[20] It is also quite likely that the name Shangdu may have come into informal use some time before 1263, so that its appearance at earlier dates is not necessarily the anachronism that Pelliot considered it to be.[21]

It seems very likely that Marco’s ‘Taianfu’, for ‘Taiyuanfu’ 太原府, may have been derived, not from any Persian form of the name, but from a Mongolian form. In Mongolian, the ‘Da Yuan’ 大元 [dynasty] was called ‘Dai Ön’.[22] ‘Taiyuan’ would therefore probably have been ‘Tai Ön’, or something very similar, hence Marco’s ‘Taian’. The origin of Marco’s ‘Pianfu’ is less clear. It may simply be that it is an early scribal error (possibly even an error by Rustichello, when first writing down Marco’s book), for ‘Pinianfu’. At least once, Rashīd al-Dīn gives ‘Tai Wan[g] Fu’ for Taiyuanfu, and transcribes ‘Pingyangfu’ 平陽府 as ‘Pung Yang Fu’.[23] So, in reality, Marco’s versions of these names do not agree with Rashīd’s Persian forms. In another place, Rashīd al-Dīn does give ‘Tayanfu’, but it is not entirely certain that this is to be identified with Tai-yuanfu. It appears that the town intended should not, in fact, be Taiyuanfu.[24] As for ‘Pianfu’, Pelliot could only say that this version of Pingyangfu ‘must be the form used by Persians in China’, without any supporting evidence.[25] Such a speculative assumption can scarcely be accepted.

‘Quengianfu’ (modern Xi’an) I have previously taken to derive from ‘Chang’an’ 長安, because it surely could not be derived from ‘Jingzhao’ 京兆, as has often been asserted.[26] The final ‘n’ of the second syllable (which Pelliot found to be unexpected and puzzling), as well as the difference of vowel in the first syllable, really make this impossible.[27] A derivation from Chang’an is also unsatisfactory, however, as ‘Quen’ cannot readily be related to ‘Chang’. I now have a much better explanation. The name is a version of Xianyang 咸陽. Later Han pronunciation of Xianyang was ‘Gεm-jɑŋ’.[28] This is so close to ‘Kinjan’,[29] or ‘Kenjan’,[30] that it must certainly be the basis of the Persian form of the name. Arabic texts of the ninth and tenth centuries use ‘Ḥumdān’ for Chang’an, very probably from the same origin.[31] Sogdians used ‘Khumdan’.[32] Xianyang, the name of the capital of the Qin 秦 dynasty (221–207 BCE), must have become known outside China in Iranian areas (certainly Sogdiana, and perhaps also Bactria and Parthia), and have continued to be used to refer to the new Han 漢 dynasty capital of Chang’an, which was very near the site of Qin Xianyang. The name then remained in use in Persian and Arabic, right up to the Yuan period. ‘Fu’ (superior prefecture) must have been added to Kenjan because of the administrative status of the city under the Song, Jin and Yuan. It is likely that there was a significant Persian-speaking colony in Jingzhao during the Yuan period. Mangghala, a son of Qubilai Qa’an, was made Prince of Anxi (Anxi Wang 安西王), with his appanage at Jingzhao, in 1272.[33] His son, Ananda, succeeded him as Prince of Anxi.[34] According to Rashīd al-Dīn, Ananda had a Muslim wet nurse, and became enthusiastically Muslim as a result, later converting most of his army of 150 thousand Mongols.[35] Muslim influence in the area of Jingzhao may therefore have been strong. No doubt at least some of the Muslims were Persian speakers. Their influence may explain why Marco uses a Persian name for this town.

Another clearly Persian name used by Marco, but which Wood does not include in her list, is ‘Çardandan [Zardandan]’. This is a translation of the Chinese ‘Jin Chi’ 金齒, meaning ‘Gold Teeth’, a name applied to a people of the Yuan province of Yunnan, and to the area where they lived[36] (today partly in Yunnan and partly in Myanmar). The custom of covering some of the front teeth with gold persisted until modern times.[37] Again, there may be an explanation for this usage. The governor of Yunnan province from 1274 until 1279 was a Muslim, Sayyid Ajall Shams al-Dīn, whose family came from Bukhara. His son, Nasir al-Dīn, who is mentioned by Marco,[38] also held various high offices in Yunnan, until 1291, when he was moved to Shaanxi.[39] It is not unlikely that the Muslim influence in Yunnan may have given currency to a certain amount of Persian nomenclature. Otherwise Marco uses very few definitely Persian terms. There is Pulisanghin, which is his name for the ‘Marco Polo Bridge’, (Lugou Qiao 盧溝橋), today in south-west Beijing.[40] He also refers to the emperor of the Southern Song as ‘facfur’. It seems that, as in the case of Kenjan, this usage went back several centuries. Versions of the word occurred in Arabic as well as in Persian.[41]

Now I have to return to Professor Morgan’s arguments. He cites Pelliot, as quoted by Cordier, as the source for:

two especially eloquent examples of ‘Persianisation.’ One is that Marco uses as his term for south China the word manzi, which is what Rashīd al-Dīn calls it but is not the Mongolian word, which is nangias … . The other example, a particularly telling piece of evidence, is that when he is discussing the Chinese/Mongolian twelve-year animal cycle, he gives ‘lion’ in place of the correct ‘tiger’. Here he cannot be translating from Turkic or Mongolian, in which the two animals are clearly distinguished. The obvious solution is that he is translating the Persian word shir, which, notoriously, can mean either ‘lion’ or ‘tiger’ ... .[42]

None of this makes any sense at all to me. I cannot understand how the fact that Marco uses a Chinese word for southern China (the area of the former Southern Song empire) can be interpreted as meaning that he was influenced by Persian usage. I would suggest that, in reality, it implies no such thing, but rather offers very slight support for the idea that he had some knowledge of Chinese. I have already dealt with the question of his use of ‘lion’ for ‘tiger’.[43] It seems, however, that I have failed to convince Professor Morgan,[44] so I shall return to this question shortly.

Morgan tells his readers that manzi 蠻子 ‘is a term which, it seems, first begins to appear in the Mongol period’, which is ‘distinctly vernacular (as is revealed by the nominative suffix zi)’ and is pejorative.[45] This is mostly correct, although I would question the use of the word ‘nominative’ (but this is not the place to discuss the peculiarities of Chinese grammar). Certainly the term is pejorative, for the character for man is written with an element meaning ‘insect’.[46] Manzi can be traced back to well before the Mongol period, however. It is the fact that it is indeed vernacular that makes it hard to know exactly when it came into use, for written Chinese was generally very different from the spoken language, at least until recent times. The ‘nominative suffix zi’ would normally not appear in written texts. The single character man is ancient, and was quite commonly used several centuries BCE. It means a ‘southern barbarian’ — that is, a person from south of the main Chinese culture area who was considered, by the Chinese, to be an inferior savage. Manzi has exactly the same meaning, but is colloquial. I have been able to trace an occurrence of the term in a work dating from about 950, where it apparently refers to ‘southern barbarians’ during the period of the Tang dynasty (the text is unfortunately defective, with missing characters, close to this occurrence).[47]

This is not precisely the meaning that it acquired a couple of centuries later, however, when it came to be used to refer to the Chinese of the Southern Song empire. This usage seems to have originated when most of China was divided between the Jin empire, founded by the Jurchens (ancestors of the later Manchus) in the north, and the Song empire in the south. There is at least one example of its use by Jin subjects to refer to southern Chinese. In 1221, when Jin armies invaded the Song empire, the Song town of Qizhou 蘄州 (in modern Hubei province, just north of the Yangtze River) came under siege. An eyewitness account, written by one of the Song defenders, describes how some of the Song crossbowmen fired at the Jin attackers bolts that were small, but could be lethal, because they were poisoned. When the Jin soldiers saw these small crossbow bolts, they cried: ‘The Manzi are firing chopsticks at us!’[48] It seems likely that it became common for northerners from the Jin empire to refer, pejoratively, to southerners from the Song empire as Manzi. This then continued during the Yuan period. This is not in any sense a Persian usage, however.

In fact, Rashīd al-Dīn does not exclusively use the term Manzi. At least three different names for southern China can be found in his writings. He uses the Mongolian word, ‘Nangiyas’,[49] several times,[50] and a third expression, ‘Machin’.[51] Indeed, at one point, he refers to ‘Machin — which the Cathaians call Manzi and the Mongols call Nankiyas’.[52] So, whatever term Marco had chosen to use for southern China, he could probably have been said to have been following Persian usage. Similarly, Juvainī refers both to ‘Khitai’ (north China) and ‘Manzi’,[53] and also, apparently with the same meanings, to ‘Chin and Machin’.[54] Jūzjānī does not appear to know the term Manzi; only ‘Chin and Machin’ appear in the Tabakāt-i-Nāsirī.[55] Mustawfī refers to ‘Khansay, capital of Machin’, and to ‘Machin. A great and extensive kingdom which the Mongols know as Nangiyas’.[56] He does not seem to have been familiar with the term Manzi. It is, therefore, apparent that Persian authors, other than those who, like Juvainī and Rashīd al-Dīn, had more extensive knowledge of the eastern part of the Mongol empire, did not use the word Manzi. If there was a distinctively Persian term for southern China, then it was Machin. This word does not appear in Marco Polo’s book.

The essentially Chinese nature of Marco’s usage is emphasised by the fact that his transcription of Manzi is a very accurate representation of the standard Chinese pronunciation of the word in his time. It must be realised that the transliteration Manzi represents modern pronunciation (in the system known as Pinyin). Marco (or Rustichello, for Marco) wrote it as ‘Mangi’.[57] The standard pronunciation during the Yuan period, as represented by the phonetic transcriptions of Chinese in the 'Phags-pa script, was man-dzhi.[58] It would, therefore, seem reasonable to conclude that Marco took this word directly from Chinese. There is absolutely no reason to think that Persian influence was involved in any way.

The question of Marco’s use of the word ‘lion’, for what were clearly tigers, is a similar red herring. I have already suggested that it must reflect the fact that neither Marco nor Rustichello knew a word for the tiger, because it was an animal that was virtually unknown to mediaeval Europeans. I have additionally pointed out that Odoric of Pordenone similarly refers to ‘lions’ in China.[59] In fact, Odoric also says that ‘black lions in very great numbers’ are found in India.[60] These must certainly have been black panthers, a colour variant of the leopard. Odoric’s usage of the word suggests that mediaeval Europeans may often have used ‘lion’ simply as a general word for any big cat. What we have here, it seems to me, is another example of the anachronistic approach to Marco Polo that has so often confused studies of his book.[61] Today, everyone knows what a tiger is, so it seems strange that someone would not recognise that an animal similar to a lion, but with stripes, was a tiger (at one point in his book, Marco actually describes a striped tiger).[62] This completely ignores the fact that mediaeval Europeans may very well have had no clear idea of what kind of animal a tiger was.

The bestiary was a popular genre of work in mediaeval Europe, especially during the twelfth and thirteenth centuries.[63] Mediaeval bestiaries quite often describe ‘tigers’. It is very obvious that their authors had never seen a tiger and had very little idea what one was like. Tigers are usually described as having spots, and being able to run very fast, which would seem to indicate that the word ‘tiger’ was actually applied to the cheetah. Many bestiaries are illustrated, and the pictures commonly show a spotted animal, with no sign of any stripes. A manuscript now in the Bibliothèque Municipale de Douai (MS 0711), in France, includes a depiction of a rather dog-like ‘tiger’, spotted all over with bicoloured rosettes.[64] This manuscript dates from the third quarter of the thirteenth century, and is therefore exactly contemporary with Marco Polo. Another bestiary, dating from somewhat earlier (about 1210), now in the British Library (Royal 12 C XIX), shows a more cat-like animal, but also spotted, not striped.[65] Both of these bestiaries, and others of the period, tell how young ‘tigers’ can be stolen from their mothers by means of a trick. The thief, although mounted on a horse, cannot move fast enough to escape the mother tiger. When she comes close, he drops a mirror, or a glass ball, and she then stops to look at her reflection, thinking it is her cub. By doing this, sometimes repeatedly, the thief is able to make good his escape.[66] Some earlier bestiaries contain even more bizarre descriptions, so that the ‘tiger’ becomes more or less a mythical beast. In some twelfth-century bestiaries, ‘a tiger is described as a kind of serpent, and is actually drawn as a dragon with wings’.[67] Neither Marco nor Rustichello would therefore have been likely to identify the striped ‘lions’ that Marco saw in China as tigers. Once again, any Persian influence can be discounted.

The list of terms used by Marco, when he is describing the Yuan empire, that are reasonably clearly of Persian origin is, then, restricted to about half a dozen. To me, at least, this hardly indicates a strong Persian influence on Marco Polo. It might just as easily be argued that he was strongly influenced by Mongolian, Turkic, or, indeed, Chinese usages, for he gives place-names in China, as well as other words, in all these languages. Indeed, I would suggest that there is better evidence that he had some knowledge of Chinese and of Turkic than there is for his knowledge of Persian. Again, I have already covered some of this ground elsewhere,[68] but it may be useful to go over at least part of it again, with additional detail. For Marco’s probable knowledge of Turkic, there is not only the evidence of a few place-names in China that are given in his book in their Turkic form, but also other Turkic vocabulary. Perhaps most tellingly, there is what Marco says about the Turkic language of the ‘Turcomans’ of Anatolia: ‘they are ignorant people and have a barbarous language’.[69] It seems at least possible that he was here contrasting the Turkic of the nomadic Turks of Anatolia with the, no doubt more sophisticated, Turkic of the Uighurs, who had become settled town-dwellers, engaged in agriculture, with a script for their language, well before the 1200s. They even made considerable use of printing.[70] For Marco to be able to make such a comparison would, of course, require him to have a good knowledge of Turkic.

One interesting piece of evidence, that suggests the importance of Turkic in Yuan China, is Marco’s use of a Turkic name for the town of Zhending 真定 (now Zhengding 正定, in Hebei province). He calls it ‘Achbaluch’, the ‘White City’.[71] What makes this especially interesting is that Rashīd al-Dīn calls the same place ‘Chagha’an Balghasun’, which is the Mongolian for ‘White City’.[72] This suggests that Marco may have heard the Turkic name more frequently than the Mongolian, although both must have been current. Certainly Marco’s use of the Turkic name does not agree with Rashīd al-Dīn’s usage. Zhending was the personal appanage of Sorqaqtani Beki, mother of two Qa’ans, Möngke and Qubilai, and of the first Ilkhan, Hülegü.[73] A number of officials appointed to office in Zhending were Turkic. They included the Uighurs Buyruq Qaya (during the reign of Ögödei Qa’an),[74] Mungsuz,[75] and Shiban.[76] The two latter were very influential under Qubilai Qa’an, so perhaps it is not surprising that Marco picked up a Turkic name for Zhending. Marco also mentions a second ‘Acbalec’, near the border between north and south China, which is distinguished by having ‘Mangi’ suffixed to it, meaning, as is explained in his book, ‘the white city of the border of Mangi’.[77] Again, Pelliot here sees a Persian construction, ‘Aq-balïq-i-Manzi’,[78] but this seems odd, as ‘Aq-balïq’ is certainly Turkic, not Persian, while ‘Manzi’, as already seen, is Chinese. It might just as well be an Italianate, or Frankish, or perhaps Latin, construction: ‘Acbalec [di/de] Mangi’. ‘Mangi’ might perhaps have been used as if it were a Latin genitive.

Another very obvious use of a Turkic name by Marco Polo is, of course, ‘Cambaluc’, from the Turkic Qan Balïkh, the ‘city of the Qan’. Pelliot notes that it is ‘purely Turkish’, but indicates that it was borrowed into Persian.[79] One might wonder why, if Persian was such an important language in the Yuan empire, the city acquired a Turkic name, which was then borrowed into Persian, rather than vice versa. Marco also gives the Chinese or Mongolian name of the same city, ‘Taidu’.[80] Pelliot says that the Chinese character for the first syllable of this name, now pronounced da, ‘was still pronounced tai [dai in Pinyin] during the Yuan dynasty’.[81] It seems, however, that this is not absolutely correct, for both pronunciations, da and dai, were probably current.[82] This use of a Turkic name for the principal capital of the Yuan empire, and (in all cases but one) of a Chinese name for the summer capital, is surely significant. If Persian had really been an important lingua franca in the Yuan empire, and had strongly influenced Marco’s use of names, then it might have been expected that some sign of this would have appeared when he was talking about the chief cities of the empire. Yet this is not the case. There is only the single use of ‘Chemeinfu’ for Kaipingfu, that is, Shangdu, to indicate some small Persian influence. This might be taken to suggest that, if Persians did have some influence in the Yuan empire, then it was certainly not paramount in the capital cities. As has been seen above, the administration of the appanage of Qubilai Qa’an’s mother had been partly staffed by Turks. Indeed, Sorqaqtani Beki was herself a Kereyid, and the Kereyids were, in all likelihood, at least partly Turkic.[83] It seems very probable that, for the Mongols, the most important group of non-Mongols were the Turks.[84] Indeed, the Yuan imperial family had marriage relations with Uighurs.[85]

Marco also uses some other Turkic vocabulary in his book. One example is bagherlac, sandgrouse.[86] Pelliot correctly notes that the form seen in most manuscripts, which puts an r after the first a, is erroneous.[87] Clauson gives bağırlak,[88] which is extremely close to Marco’s version of the word. The particular bird that the word is applied to is almost certainly Pallas’s Sandgrouse.[89] Another word of Turkic origin, but apparently borrowed into Mongolian, is ‘toscaor’ or ‘toscaol’, ‘men who stay to watch’, or, more simply, watchmen.[90] Even Pelliot, although he found a related usage in Rashīd al-Dīn, did not consider the word, in Marco’s form, to show any Persian influence.[91]

A further instance, of considerable interest, is Marco’s use of the term ‘Argon’, which he defines as meaning ‘guasmul in French, that is to say that they are born of the two races, of the lineage of those of Tenduc who worship idols and of those who worship by the law of Mahomet’.[92] Pelliot explains that ‘guasmul’ means someone, ‘one of whose parents was Latin and the other Greek’.[93] Arkun, in mediaeval Turkic, had the meaning ‘cross-bred’,[94] so the correspondence is clear. According to Pelliot, however, Argun was also a tribal name, of a people apparently from a region between Talas and Balasaghun. He suggests that a substantial number of these people were moved eastwards by the Mongols to the region where Marco Polo came across the Argon, and that Marco heard ‘the name of the Mussulman Arγun settlers; but he knew also the Turkish word arγun used in the sense of half-breed, and applied it wrongly’.[95] If this is correct, then it is further evidence that Marco had at least some knowledge of Turkic.

Marco also uses various Mongolian words, of course. Some were common to both Mongolian and Turkic. They include titles, the name of an animal, and other vocabulary. Thus, he refers to ‘Baian Cingsan’,[96] the conqueror of Mangi. This is Bayan of the Bārin, and the title is the Mongolian form, chingsang,[97] of the Chinese chengxiang 丞相, Chancellor, a title which Bayan did indeed hold.[98] Marco also refers to the rebellious Chinese warlord, Li Tan 李璮, as ‘Liitan Sangon’.[99] Pelliot’s explanation of this, which, as he himself noted, differs from that of most other commentators, is quite wrong. He rejects the Mongolian (and Turkic) form, sänggün or sänggüm, of the Chinese jiangjun 將軍 (military) General, as the origin of ‘Sangon’, claiming that jiangjun was ‘rarely used’ during the Mongol period. Instead, he suggests that xianggong 相公, ‘Duke minister’, which ‘in the Middle Ages was applied to young men of high families’, was the Chinese original of this term.[100] Yet a search of the Yuan shi reveals that jiangjun appears quite literally hundreds of times, while xianggong is extremely rarely used, with only four occurrences that I have been able to trace. Jiangjun also occurs more than a hundred times in the Guochao wenlei 國朝文纇, while xianggong occurs only twelve.[101] Pelliot’s theory can therefore be rejected. ‘Sangon’ certainly derives from sänggün, which in turn was derived from the Chinese jiangjun.

Another title appears in Marco’s ‘Vonsamcin’, one of the commanders of the second Mongol invasion of Japan.[102] The person intended was Fan Wenhu 範文虎, but ‘samcin’ comes not from his name but from his title. In Chinese, this was canzhi zhengshi 参知政事 (Second Privy Councillor), commonly shortened to canzheng 参政. The Mongolian equivalents were samji jingshi and samjing.[103] Clearly, Vonsamcin is Marco’s version of Fan Samjing. The Yuan shi records that Fan Wenhu was promoted from Second Privy Councillor to be a joint Junior Vice Councillor of the Central Secretariat in 1278.[104] Other Mongolian words used by Marco include ‘quesitan’,[105] from the Mongolian kešigten, members of the kešig, or personal guard of the Qa’an;[106] ‘gudderi’,[107] meaning the musk deer;[108] ‘burcan’,[109] which is burqan, the Mongolian (and Turkic) word for Buddha;[110] and, rather obviously, the titles of the Mongol rulers, ‘Kan [Qan]’ and ‘Kaan [Qa’an]’,[111] which are of ancient usage, both in Turkic and in Mongolian.[112] There is, therefore, far more Mongolian vocabulary in Marco Polo’s account of the Yuan empire than there is Persian.

In fact, there is very little evidence to support the idea that Marco Polo could understand and speak Persian. I do not wish to question this belief, as it seems probable that he did learn Persian. Nevertheless, to my mind, there is much better evidence that he had a knowledge of Turkic (as has been outlined above), and quite good evidence that he had at least an imperfect command of spoken Chinese. His probable confusion of Jin 晉 and Jin 金, when he was in Xiezhou 解州 (in modern Shanxi province), is only likely to have occurred if he had heard the name spoken, and understood it wrongly.[113] If he had been able to read Chinese characters, then such confusion would have been improbable. Perhaps, if he did indeed learn at least some Chinese, he learned only the 'Phags-pa script for written Chinese.

Morgan notes that Turkic was probably very important in the early Mongol empire, and quotes Juvainī’s ‘well-known grumble’ that the Uighur language and script had come to be considered ‘the height of knowledge and learning’ in his time. Yet Morgan immediately goes on to say that: ‘It does not seem likely that Uighur Turkish, if it seemed pre-eminent in those early decades, retained any kind of supremacy over Persian permanently’.[114] He gives absolutely no explanation of this claim, nor does he provide evidence to support it. It can only be assumed that he believes that the superiority of Persian is self-evident. It is very probably true that, in the Ilkhanate, Persian became the principal language of administration under the Mongols, as Fragner has stated.[115] It is undoubtedly the case, however, that Persian had no claim to any kind of superior status in the Yuan empire, where Chinese culture and the Chinese language and script were predominant. To the Chinese, Persian would have been just another ‘barbarian’ language. Indeed, during the Mongol/Yuan period, the Chinese (and very probably also the Mongols) did not even distinguish clearly between Persians, other Muslims, and the various other peoples from the ‘Western Regions’ (Xi Yu 西域).[116]

Professor Charles Melville has expressed a view quite opposite to Morgan’s: ‘Juvainī’s complaint that a knowledge of the Uighur script was a passport to advancement was probably as true at the end of the Ilkhanate as at the beginning.’[117]

The pre-eminent status of the Uighurs was clearly stated by Qubilai Qa’an himself in 1270. When the Koreans complained that they were accorded a lower status than the Uighurs at the Yuan imperial court, Qubilai replied:

You submitted later, therefore you are ranked low among the princes. During the reign of our Tai Zu [Chinggis Qan], the Iduq qut[118] was the first to submit. Therefore it was ordered that he be ranked first among the princes. Arslan[119] submitted afterwards, therefore he was ranked below him. You should know this![120]

The Koreans no doubt felt that they were a civilised people, with a long history of relations with China, so that they deserved to rank above the Uighurs. Probably the Persians felt similarly about Turks. Qubilai, however, quickly put the Koreans in their place. The idea that Persian would surely soon have replaced Turkic, presumably because of its cultural superiority, can therefore be dismissed.

Morgan quotes Leonardo Olschki on the status of Persian as a language ‘commonly used … for purposes of business and trade’.[121] Olschki also says, however, that, in the fourteenth century, ‘the Turkish dialect of the Comans … was, … together with Persian, the language spoken or understood throughout the Tartar empire from Persia to Cathay’. He goes on to opine that Turkic ‘certainly must have been one of the various languages Marco claims to have learned’.[122] This is surely correct, as I have already suggested above, for by the thirteenth century Turkic, at least as much as Persian, had become commonly used as a language of commerce, right across Central Asia.[123] Today, various Turkic languages are spoken across a broad stretch of the Asian continent, from Xinjiang in the east to Turkey in the west. Turkic languages have more or less completely replaced all the earlier languages of this region. Of the inhabitants of Central Asia today, 89 per cent are Turkic speakers.[124] At the time of the Mongol conquests, the process of Turkicisation was already well advanced (indeed, in the west, in the area near the Black Sea, Turkic peoples and languages were probably more widespread than they are today).[125] The mixing of Iranian peoples, such as the Sogdians, with the Turks had a long history by the year 1200. For example, An Lushan 安禄山 (703–757), whose rebellion in 755 almost caused the collapse of the Tang dynasty, was the son of a Sogdian father and a Turkic mother.[126] Sogdian documents from Mount Mugh, dating from 710, mention a bridegroom with a clearly Turkic name (‘Ot-tegin’) marrying a probably Sogdian wife.[127] There was Turkic influence on the Sogdian language, from as early as the time of the trilingual Qarabalghasun inscription (early 800s). During the ninth to tenth centuries, a language often called ‘Turco-Sogdian’ had developed, that is, Sogdian showing strong Old Turkic influence.[128] By the time the Mongols invaded Sogdiana (then part of the Khwarazmian empire, which had recently taken it from the Qara Khitayans),[129] it was very likely at least as much a Turkic as an Iranian area.[130] Indeed, much of Khwarazm itself seems already to have been largely Turkicised. After travelling across Central Asia in the mid-1240s, John of Plano Carpini said: ‘On leaving the land of the Kangits [Qangli] we entered the country of the Bisermins. These people used to speak the Coman language, and still speak it, but they hold the faith of the Saracens.’[131] The ‘Bisermins’ were the Khwarazmians.[132] In relation to this country, John goes on to relate that, on a large river of which he did not know the name, there were cities called ‘Iankinc’, ‘Barchin’ (or ‘Karachin’), and ‘Orpar’ (or ‘Ornas’).[133] The first is surely Yangikent, so that the river would be the Syr-Darya. Orpar has already been identified as probably being Otrar,[134] which is indeed on the same river. ‘Barchin’ has been identified as Barjligh-Kent, said to be ‘somewhere between Jand and Sughnaq’,[135] or, if ‘Karachin’ is assumed to be the more correct reading, it might possibly be Khojend. In either case, it would be on or very near the Syr-Darya.[136] It would seem, then, that by the 1240s, when John passed through it, the area around the Syr-Darya was already Turkicised.[137]

At a slightly later date, William of Rubruck made a similar journey to that of John’s, apparently passing south of Lake Balkhash and spending twelve days in Qayalïq (if his ‘Cailac’ is this town, as seems very likely). He records that the area around the town ‘used to be called Organum and used to have its own language and script but now it has all been seized by the Turcomans’. He also says that there was a market in Qayalïq, to which ‘many merchants flocked’.[138] Presumably, the principal language of this market would have been Turkic, of some form. The resemblance of ‘Organum’ to Arkun or Arγun (already discussed above) is striking. Perhaps the word was used in this area, too, for people of mixed descent, partly Turkic, interbred with the earlier inhabitants. As Morgan notes, William also mentions ‘a fine town called Equius, inhabited by Saracens who spoke Persian, though they were a very long way from Persia’.[139] As this town is mentioned before Cailac, when William was travelling eastwards, it should presumably be west of it. This would rule out the identification with Quyas, which Morgan embraces, following Pelliot (although Pelliot also considers another possibility, Iki-ögüz).[140] A major question here is what exactly William meant by ‘Persian’ and ‘Persia’.[141] There were peoples at least not very far removed from the area between Talas and Qayalïq, where ‘Equius’ apparently was situated, who spoke Iranian languages. There may very well have been Sogdian settlements in the area, as there had also been much further east, at one time, and Sogdian is an Iranian language.[142] Perhaps, then, William’s ‘Persian’ was Sogdian, in which case, these Persian-speaking Saracens were not so very far from home.[143] They may well have been simply a surviving remnant, isolated among the Turkic incomers, of the earlier inhabitants of the region.[144] Nevertheless, what William says here about being ‘a very long way from Persia’ is significant. Clearly, to him, Persia was distant from the location of ‘Equius’. It would seem, then, that this entire region was clearly not Persian when William passed through it. The fact that he found some Persian speakers there was evidently unusual and worthy of comment.

It is also worth noting that Marco Polo gives an account of what he calls ‘the great Turquie’, by which he apparently means Turkestan: ‘The great Turquie is beyond the river of Gion and stretches from toward tramontaine as far as to the lands of the Great Kaan’.[145] The ‘Gion’ or Gihon was, of course, a name often used at the period for the Amu Darya. Thus, it would seem that all the lands from the Amu Darya as far as the boundary of the empire of Qubilai Qa’an was ‘Turquie’ to Marco Polo. He also refers to Qaidu going into ‘the great Turquie to Samarcan’,[146] indicating that Samarkand (and presumably all of Transoxania) was in ‘Turquie’. Although he himself had probably never been to Samarkand, his father and uncle had very likely been there, and had certainly spent time in Bukhara,[147] so his testimony is likely to be reliable.

On the question of the languages used in trade during the Mongol period, it is probably not without significance that the word adopted by the Mongols for the ‘merchant partner’ system was of Turkic origin. Ortaq (borrowed into Mongolian as ortogh) is a Turkic word meaning ‘partner’.[148] Morgan notes that: ‘These Muslim merchants [of the ortogh] were no doubt of very varied ethnicity … . At least they will have spoken a wide variety of languages …’.[149] This is probably generally correct, but it must be likely that Turkic languages were predominant, at least on the overland trade routes through the largely Uighur regions around the Tarim Basin, and the mainly Turkic Semirechye and Khwarazm. If Persian were really the principal lingua franca of trade, why was a Turkic name adopted for the ortaq system, rather than a Persian one? It appears that New Persian never penetrated far along the ‘Silk Road’. Of the documents from Chinese Turkestan available in the database of the International Dunhuang Project, only nineteen fragments are in the New Persian language, and written, not with Arabic, but with Syriac script. All of them originated from Turfan.[150] Probably they are documents of Nestorian Christian (or perhaps Manichaean) origin. The Qarakhanids, who controlled a large part of Central Asia during the eleventh and early twelfth centuries (including Transoxania), ‘cherished their Turkish ways’, and ‘fostered the development of a new Turkish literature alongside the Persian and Arabic literatures that had arisen earlier’.[151] On their coins, Arabic inscriptions predominated, but mixed with both Turkic and Persian.[152]

Morgan also raises the issue of the language used in communications between the Mongols and the Papacy. He quotes John of Plano Carpini at some length, regarding the languages used for writing the letter which John carried from Güyük Qa’an to the Pope. Impressively, he reveals that the Persian original of the letter still exists in the Vatican archives, bearing the seal of Güyük Qa’an.[153] It seems to me, however, that a language used for a diplomatic communication with a foreign dignitary need not have been one that was any kind of lingua franca in the place of issue. The question was whether the Pope would be able to find, in the area where he resided, someone who could read the letter. John makes it quite clear that original versions of the letter were carefully prepared in more than one language, including Latin.[154]

He says that: ‘we were asked if there were any people with the Lord Pope who understood the writing of the Russians or Saracens or even of the Tartars’. Clearly then, one option was to send a letter written in Russian (or perhaps, at this period, it would be better to say Slavonic). I do not believe anyone has ever suggested that Russian was a lingua franca in the Mongol empire. John responded that ‘there were … Saracens in the country but they were a long way from the Lord Pope’. It is not very clear exactly what he may have meant here, but probably there were Turks who were closer to the Pope than any Persians. There were certainly Turks in Anatolia at this period, and even in the Balkans, besides those with whom the Crusader states had contact.[155] Some of the Cumans had converted to Latin Christianity in the 1220s,[156] so that there were close contacts between the Roman Church and Turks.

The fundamental issue here, is what John meant by ‘the writing of … the Saracens’. Morgan seems convinced that it must have meant Persian, but this is by no means certain. It may be noted that John in fact refers to Saracen writing in a different section of his narrative, which provides a useful comparison: ‘we delivered the letter [from the Pope] and asked to be given interpreters capable of translating it. We were given them on Good Friday, and carefully translated the letter with them into Ruthenian, Saracenic and Tartar characters.’[157] This occurred when he was at the orda of Batu, somewhere near the River Ural. It seems unlikely that ‘Saracenic’ here would mean Persian (although ‘Saracenic … characters’ may have meant the Perso-Arabic script, used to write Turkic). Far more likely is that John here meant Russian (or Slavonic), Turkic and Mongolian. Batu, after all, was ruler of the Qanate of Qipchaq, where Turkic (the Cuman language) was predominant. According to Fragner, while Persian became the principal language of administration in the Ilkhanate, the rulers of the Jochid Ulus ‘preferred Qipchaq Turki’.[158] The ambiguity of the word ‘Saracen’ is shown by what William of Rubruck says about the ‘Iron Gate’ (in the Caucasus), ‘which is on the route of all the Saracens coming from Persia and Turkey’.[159] So Saracens could clearly be either Turks or Persians. Also of significance is that William says that, during his return journey, Batu assigned him a guide who was an Uighur.[160] Again, if Persian was such an important lingua franca, why give William a guide who was a Turk? William clearly states that he ‘believed that you [King Louis IX of France] were still in Syria and directed my journey towards Persia’.[161] So he was given a Turkic guide to lead him towards Persia! It may be worth noting here that Persia had itself been under Turkic rule for a long time before the period of the Mongol conquests. The Seljuqs, who were, of course, Turks, had conquered Iran in the mid-eleventh century, and ruled there until about 1200. The chronicler Matthew of Edessa, writing in the 1130s, frequently refers to ‘Turks’ and ‘Persians’, without any discrimination between them. For example, he relates that:

In the year 571 of the Armenian era [1122–23] the Persian general Il-Ghāzī collected troops and marched against the Frankish forces. First he descended upon Aleppo and from there went and encamped in the Muslim town of Shaizār. Baldwin, the king of Jerusalem, came and was joined by the count of Edessa Joscelin; then both marched forth and encamped opposite the Turkish forces.[162]

Subsequently, the Turkic Khwarazmshahs Tekish and his son Qutb al-Dīn Muhammad took control of most of Iran, until the Mongols destroyed the Khwarazmian empire.[163] Persia had, therefore, been under considerable Turkic influence for more than a century and a half before the period of the Mongol conquests. Indeed, it has been suggested that the New Persian language developed at least partly under Turkic influence.[164] It is commonly overlooked that Turkic is widely spoken in Iran; it has been suggested that there may, in fact, be more speakers of Turkic in Iran than of Persian. ‘It is generally thought that in the land of Persia, nothing is spoken but Persian, and few are aware that Turkish is widespread throughout Iran. It is perhaps even more common than Persian …’.[165] In this context, Judith Pfeiffer has stated that: ‘Mongol rule in the Middle East extended westwards until the Euphrates River, eventually resulting in the Euphrates becoming a political and cultural border zone, with an Arabophone zone south of the river, and a Perso-Turkish zone to the north of it’.[166]

An indication that the Persian letter in the Vatican archives may have been translated from a Turkic original is that its opening phrase, which Pelliot translates: ‘Dans la force du Ciel éternel, [nous] le Khan océanique du grand peuple entier; notre ordre’, is actually in Turkic (written with the Perso-Arabic script). It corresponds more or less exactly with the Mongolian wording of the Qa’an’s seal on the document.[167] It seems most unlikely that, if the original of the letter had been entirely in Persian, then its opening phrase would have been translated into Turkic. On the other hand, if the original had been in Turkic, then it is conceivable that this opening phrase might have been left untranslated, as it was a standard formula (as evidenced by the fact that it was engraved in Mongolian on the Qa’an’s seal). Whatever the case, it seems that the Turkic version of this formula was so familiar to whoever originally wrote the letter that he wrote it in Turkic, even though the rest of the text was Persian. Presumably, then, he was not familiar with a Persian version of the formula. Turkic, therefore, must have been more commonly used than Persian at the court of the Qa’an. This is also an interesting early use of the Perso-Arabic script for writing Turkic.

A final piece of evidence in favour of the ‘Saracenic’ used for writing to the Pope being Turkic is that Marco Polo clearly states that: ‘the great lord [Qubilai Qa’an] had his letters and privileges immediately made in the Turkish tongue to send to the Apostle [the Pope] and entrusts them to the two brothers [Marco’s father and uncle] and to his baron …’.[168] Here, then, is an unequivocal statement that letters to the Pope were written in Turkish. Moreover, this was in or about 1267,[169] after Turkic should have begun to be replaced by Persian, according to Professor Morgan. Even if this story is untrue (as some would probably argue), it nevertheless demonstrates that, during the late thirteenth century (when Marco’s book was written), Turkic could have been thought to be an appropriate language for a Mongol Qa’an to use when writing to the Pope.

It is also worth noting that the Chinese term Huihui zi 回回字, which might be considered equivalent to ‘Saracenic characters’, can mean the Uighur script. A Chinese work dating from 1237 says the following regarding Mongol writing:

The Tatars [Mongols] originally had no script or documents. However, there are three kinds that they now use. That which is current in the original country of the Tatars is only to use small pieces of wood three or four inches long, cut at the four corners. Thus, if ten horses are concerned, then ten cuts are made. In general, only the number is cut. … These small pieces of wood are like the old tallies [of China]. That which is current among the Huihui 回回 uses Huihui script. Chinqai[170] is in charge of it. There are only 21 letters in the Huihui script.[171] ... That which is used in the lost [conquered] countries of the north Chinese, the Khitans and the Jurchens is only Chinese characters.[172]

Thus, it would appear that only the Uighur script and Chinese characters were in common use among the Mongols in the 1230s. It is by no means impossible that John’s ‘Saracenic writing’ might have been Uighur script, used to write Uighur Turkic.

The claim has been made that Persian was an ‘official’ language of the Yuan court. It is repeated by Morgan, quoting Igor de Rachewiltz, who based his opinion on an article by Huang Shijian.[173] Great as is my respect for Professor de Rachewiltz, I am convinced that he is quite wrong about this. Indeed, it seems that Huang Shijian persuaded him to alter an earlier opinion, for at one time he asserted that: ‘the lingua franca of the Mongol empire, at least in its eastern portion, was almost certainly not Persian, but Turkish’.[174] Huang’s evidence for this claimed ‘official’ status of Persian is, as Morgan notes, based on ‘scattered … pieces of evidence’. Nevertheless, Morgan continues, ‘evidence for anything in the Asia of the 13th and 14th centuries’ is similarly scattered.[175] This is debatable, at least in the context of China during the thirteenth and fourteenth centuries. There is, in fact, a very great deal of evidence for the situation in China at that period. Of course, there are lacunae, but if Persian had really been an ‘official’ language of the Yuan court, then there should be much more evidence for this than the meagre fare offered by Huang. His pieces of evidence are: (1) the Chinese term Huihui, in the great majority of instances, means ‘Persian’; (2) there are (as of 1986) inscriptions in Persian on two Mongol paizi 牌子 (‘tablets of authority’, as Marco Polo called them),[176] and on some officially issued standard weights; (3) a Muslim National College (Huihui Guozi Xue 回回國子學) was established in 1289, with the principal purpose of teaching Persian.

None of this evidence is at all convincing. Firstly, the term Huihui quite certainly cannot be considered to be to any degree synonymous with ‘Persian’. It means ‘Muslim’ or ‘Islamic’, and, like the mediaeval European term ‘Saracen’, it was applied to people of various backgrounds who were Muslims, including both Persians and Turks (and also Arabs, of course: the Abbasid Caliph is referred to as Huihui Halifa 回回哈里發 in the Yuan shi).[177] Indeed, it was sometimes used even more loosely, to refer to Jews (Zhuhu Huihui术忽回回) and even to some groups of Christians.[178] It is in fact quite likely that the majority, or at least the largest single group, of Muslims in China were Turks. Indeed, the term Huihui is a variant of the Chinese transcription of the ethnonym ‘Uighur’. This was originally Huihe 回纥 or Huihu 回鶻, but during (or even before) the Yuan period it became Huihui, and was applied to Muslims in general. Presumably this was because the first large group of Muslims with which the Chinese became familiar was the Uighurs (although it must be noted that they were not all Muslims during the Yuan period; it was the Uighurs of the eastern part of the Qarakhanid realm — the oases of the western Tarim Basin — who were probably the Muslim Uighurs known to the Chinese).[179] Indeed, Huihui was sometimes used in the sense of ‘native of Turkestan’.[180] A new Chinese transcription of the name of the Uighurs came into use, Weiwuer 畏兀兒 (or sometimes just Weiwu 畏兀, as well as other variants).[181] It has already been shown above that much of Central Asia had become largely Turkicised by the period of the Mongol conquests: Khwarazm is referred to in Yuan Chinese sources as ‘the Country of the Muslims’ (Huihui Guo 回回國),[182] and a large part of Khwarazm, at least from the Semirechye to the Aral Sea, was mainly Turkic.

It is also important to note that the Mongolian equivalent of Huihui was Sarta’ul.[183] This word probably derives from a Sanskrit original, meaning ‘merchant’. It came to be applied to the settled, non-nomadic inhabitants of Turkestan and, at least by the nineteenth century, in the form Sart, ‘was used by the Russians, and by Western travellers, as a name for any Turkish-speaking, Muslim, non-nomad, oasis-dweller of Russian or Chinese Turkistan’.[184] As already seen above, Huihui was derived from the Chinese transcription of ‘Uighur’, and sometimes meant a ‘native of Turkestan’. It may well be that Sarta’ul and Huihui already, during the thirteenth and fourteenth centuries, had something of the nineteenth-century sense of Sart. Indeed, Maḥmūd Yalavach and his son Mas‘ūd are described as ‘Sarta’ul of the Qurumši clan’ in the Secret History of the Mongols.[185]

That Huihui were likely to be Turks is shown by a passage in one of Zhou Mi’s works. He gives a description of ‘How Muslims [Huihui] bury their dead’, which he says was based on what he had personally witnessed in 1291:

The custom of Muslims is that, whenever someone dies, there is a person, who specialises in washing the corpse, who pours water from the mouth of a large copper urn and washes the stomach and abdomen to get rid of all the unclean qi 氣. Then the body is washed clean from head to foot. After the washing has been completed, it is wiped dry with a cloth. Then a bag is made of ramie, silk or hempen cloth[186] and the body is put into it naked. Only then is it placed in a coffin. The coffin is made of thin pine boards and is only big enough for the corpse, nothing else at all is put into it. The dirty water from the washing of the body is collected in a pit under the room and covered with a stone: this is called ‘summoning the spirit’. They set up a table above the pit. Every four days an offering of food is made. After forty days this ends, and on a suitable day the coffin is taken out and interred in the Ju Jing Yuan 聚景園. This garden is in the charge of a Muslim. The rent of every plot of land has a regular price, and the overseer of the garden has all the bricks, mortar and labour used, which he sells for money. When it comes to the time of mourning for the dead, the relatives all cut their faces, tear their hair and rend the seams of their clothes. Staggering and wailing, they move [the hearts of everyone] near and far. When the coffin is carried out, the rich get beggars to hold candles and scatter fruit along the road; the poor do not do this. Then everyone in order, young and old, bows and kneels, as is the common custom. When the obeisances have been completed, they make a noise with the tips of their boots by way of music, and comfort each other. When they have fully expressed their feelings, they get all the Muslims to recite their sacred texts. Three days later, they again go to the place of burial. The rich mostly kill oxen and horses and give a banquet for their fellows, even down to the poor and beggars of their neighbourhood. It is also said that sometimes when the coffin arrives at the place of burial, the body is removed from it and buried naked in the grave with the face towards the West.[187]

The Ju Jing Yuan was on the shore of the West Lake at Hangzhou.[188] What is of greatest interest here, however, is that these Huihui were clearly Turks. Lacerating the face as a sign of mourning was very distinctively a Turkic custom. Killing oxen and horses for the funeral feast was also a steppe tradition, from the time when Turks were nomads (as, indeed, many still were at this period).

The Turkic custom of lacerating the face as a sign of mourning can be traced back several centuries before the Yuan period. When Attila died in 453, it was part of the mourning ritual of the Huns.[189] In 576, Byzantine ambassadors to the Türks attended the funeral of a chief, and were obliged to undertake it.[190] It is also mentioned in the Chinese annals of the Tang dynasty. A Chinese princess, who had married the Qaghan of the Uighurs, was apparently expected to commit suicide so that she could be buried with him when he died. She avoided this fate by using the excuse that her late husband, by taking her as his wife, had shown his admiration for Chinese customs, which did not include such a requirement. Nevertheless, she ‘observed Uighur custom by slashing her face and weeping loudly’.[191] A Chinese work completed in 1285 describes the mourning customs observed by the peoples from north of the Gobi:

According to custom north of the desert, when someone died, the body was placed in a tent. The sons and grandsons, and other members of the family, male and female, each killed an ox and [or?] a horse, and placed them in front of the tent as sacrificial offerings. They rode round the tent on horseback seven times. Going to the door of the tent, they slashed their faces with a knife and wept, so that blood and tears flowed together. This they did seven times.[192]

Cutting the face, weeping loudly, and killing oxen and horses would, therefore, seem to have been old Turkic customs, which had persisted even after the conversion to Islam. Zhou Mi was resident in Hangzhou at the time when he wrote his description of Islamic funeral customs, and the reference to burial in the Ju Jing Yuan makes clear that he was referring to Muslims of Hangzhou. This city, on the east coast of China, at the southern end of the Grand Canal, was the former capital of the Southern Song empire (Marco Polo’s Quinsai).[193] It was a major population centre, with a flourishing commerce, attracting sea-borne trade, and no doubt also merchants from all over the Mongol empire. Its Muslim community probably represented a good cross-section of the Muslims in China at the period. If, for Zhou Mi, Huihui were Turks, then it seems likely that Turks formed at least a substantial fraction of all Muslims in the Yuan empire, if not an outright majority.

It is also known that a number of prominent Muslims in Mongol service were, or probably were, Turks, or, at least, speakers of a Turkic language. They include Mahmud Yalavach, who was ‘a Turkish speaker from Khwarazm … and … a merchant by profession’.[194] Many of his descendants also served the Mongols, for several generations. There is no clear information about the language or ethnicity of Sayyid Ajall Shams al-Dīn, but his family came from Bukhara, in what had been Sogdiana but was strongly Turkicised by the Mongol period. It must at least be likely that he had some knowledge of Turkic, even if it was not certainly his first language.[195] The infamous Ahmad came from Fanākat (Benaket), on the Syr-Darya, near modern Tashkent in Uzbekistan, and may very well have been a Turk, or at least of mixed Iranian and Turkic stock.[196] There are also the three thousand Muslim artisans who were transported from Samarkand, Bukhara, and other places, and installed in Xunmalin 尋麻琳, near modern Zhangjiakou (formerly Kalgan).[197] Most of them were probably Turks, or at least of mixed parentage, and partly Turkic. A search of the History of the Yuan Dynasty has revealed no Muslims (Huihui) for whom there is any very definite indication of Persian origins.

There are, however, Persian inscriptions on surviving tombstones, which indicate that there must have been a significant number of Persians in China (particularly in Quanzhou and Hangzhou) during the Yuan period. Nevertheless, the surviving inscriptions are mostly in Arabic, with only a small minority in Persian. They include a number that are memorials to Muslims who were probably Turks. For example, of the gravestones from Quanzhou described by Chen Dasheng,[198] only some half a dozen bear inscriptions in Persian, all of which also bear Arabic inscriptions, while there are more than two dozen with inscriptions only in Arabic. The ethnicity of most of the deceased cannot be determined with any kind of certainty, but perhaps a dozen were probably Persians, while about half as many were Turks. In this south-eastern port city, this preponderance of Persians is scarcely surprising. During the Yuan period, Quanzhou was the principal port for commerce with South-east Asia, India and the Persian Gulf.[199] It seems quite likely that the maritime trade routes, particularly those leading towards the Persian Gulf, tended to be dominated by Persians (and no doubt also Arabs), while the overland routes, from north-western China across Central Asia, were to a great extent the preserve of Turks. For centuries, the Turks had been influenced by the Sogdians, the great traders of the overland ‘Silk Road’. They had intermarried with them (as already seen above), and eventually had submerged them under a wave of Turkic migration.[200] They had probably also replaced them as merchants.

If perhaps no more than a third of Muslims in Quanzhou were Turks, it is probable that, further to the north and west, the proportion of Turks was higher.[201] Overall, it is quite possible that a majority of Muslims in the Yuan empire were Turks, or at least speakers of a Turkic language. Since there were also many non-Muslim Turks, it is quite clear that Turks must have greatly outnumbered Persians. It has sometimes been said that the majority of the ‘Classified Peoples’ (Semu ren 色目人) were Muslims.[202] This is certainly not true. The Classified Peoples were a very diverse group, including many different peoples who were not Muslims. The Huihui were only one group among many included in this class. The idea that large numbers of Persians were among the Classified Peoples was rejected three decades ago.[203]

It should also be noted that by no means all religious inscriptions from Quanzhou are Islamic. There are also Nestorian Christian, Manichaean and Hindu relics from the Quanzhou area.[204] Most of the Nestorian ones, at least, must date from the Yuan period, as Nestorian Christianity was proscribed during the late Tang dynasty and more or less entirely disappeared from China until after the Mongol conquest.[205] The religion flourished under the Mongols, some of whom were themselves Nestorians. The mother of the Qa’ans Möngke and Qubilai, Sorqaqtani Beki, was a Nestorian Christian.[206] Many, very possibly most, of the Yuan-period Nestorians, who were by no means insignificant numerically, must have been Turks. The inscriptions on Nestorian tombstones from Quanzhou are in Syriac script, but the language used in the main body of most of them is Turkic.[207] In fact, a considerable number of Nestorian inscriptions in Turkic written with Syriac script, and sometimes with Uighur script, are now known from China. Most Nestorian remains have been found in north-west China and Inner Mongolia, but there are a significant quantity from Beijing, Yangzhou and Quanzhou.[208] It seems sometimes to be believed that most rich merchants in the Yuan empire were Muslims, but this is not necessarily so. A decree of Möngke Qa’an quoted by Wang Yun refers to ‘Uighurs and Muslims who are engaged in trade’.[209] It is quite likely that many of the Nestorian Christian Uighurs in Quanzhou were merchants. There were also wealthy Chinese merchants.[210] An inscription in Tamil discovered in Quanzhou records the installation of an image of a god in a Hindu temple in the city in 1281.[211] The temple must surely have been financed by Indian merchants trading in the port.

All people in the Yuan empire were categorised into one of four classes, of which the Semu ren formed one. The question of who the Semu ren were is perhaps best approached by first considering who they were not. They were not Mongols, of course: the Mongols were the highest-ranked group, the privileged conquerors, at the top of the social scale in the Mongol empire. At the bottom of the pile were the southerners, (Nan ren 南人), overwhelmingly Chinese, the former subjects of the Southern Song empire. Above them came the northerners (Han ren 漢人), many of whom were also Chinese, who had been subjects of the Jin empire of northern China. My translation ‘Northerners’ requires explanation. Today, Han is used with the meaning ‘ethnic Chinese’, in distinction from the various ethnic minorities of China. This was not its usage during the Yuan period, for Han ren included not just Chinese, but all the peoples who had been subjects of the Jin empire, including Jurchens and Khitans, among others. It also excluded the Chinese of the south.

It is quite often said that this classification of peoples under the Mongols was based on ethnicity or race.[212] This is obviously untrue, for ethnic Chinese were divided between two of the categories: some were Southerners, others were Northerners. Nor were they the only group to be so divided. The Yuan shi records that ‘people from the Hexi 河西 region,[213] Muslims, Uighurs, and so on’, could hold offices up to the rank of darughachi (overseer, or imperial agent) of a Myriarchy (Wanhufu萬户府), in the same way as the Mongols, whereas Jurchens and Khitans were subject to the restrictions on holding high office that were imposed on the Han ren. ‘But those Jurchens and Khitans who were born in the north-west, and who do not understand Chinese, are the same as Mongols’.[214] This presumably refers to those Khitans and Jurchens who had been subjects of the Qara Khitai empire.[215] Thus, like Chinese, who could be either Nan ren or Han ren, Jurchens and Khitans could be Han ren or Semu ren.

Other Semu ren were Tanguts, who would have been many of the ‘people from the Hexi region’ just mentioned;[216] Uighurs, and various other Turkic peoples, including Qarluqs, Qanglis, Öngüts, and Qipchaqs; Naimans (who may also have been Turks);[217] Alans from the Caucasus; Tibetans; Kashmiris, and several others. Among these were, of course, the Muslims or Huihui.[218] Basically, Semu ren were all those who were neither Mongols, nor former subjects of the Jin and Southern Song empires. These classifications were based on the perceived loyalty of the various peoples. Those who had submitted (or had been subjugated) first, like the Uighurs (as seen above), and the Khwarazmians, were considered by the Mongols to be more likely to be loyal than were the Northerners of the Jin empire, who had held out against the conquerors until 1234, and especially the Southerners of the Song empire, who had not been forced into submission until the 1270s.

It is very difficult, if not impossible, to get any clear idea of how many of each of the various groups of Semu ren there were in the Yuan empire. Indeed, there is no good figure for the total of Semu ren. A very approximate notion of the relative importance of the various Semu peoples can be gained from how often they are mentioned in the Yuan shi.[219] The term Huihui occurs about 180 times. Qipchaq (Qincha 欽察) appears more than 160 times; Uighur (Weiwu or Weiwuer), just over 40; Qangli (Kangli 康里), more than 30; Qarluq (Halalu 哈拉鲁), about ten; Tangut (Tangwu 唐兀), about 70; and Naiman (Naiman 乃蠻), more than 90. This can only give a very rough idea of relative numbers of these various Semu ren. Nevertheless, it seems clear that, although the Muslims were a major group, they were certainly not the majority of Semu ren.[220] It also seems very probable that the various Turkic peoples collectively outnumbered the Huihui. Since many, if not most, of the Muslims were also Turks, it is entirely reasonable to conclude that there were far fewer Persians than Turks in the Yuan empire.

Another approach to the same question results in a similar conclusion. In the early 1980s, Igor de Rachewiltz produced a detailed study of Turks in the Yuan empire.[221] In it, he provides counts of Turks who played a significant role in some aspect of government or culture at various times during the Yuan period. He excludes those about whom there is only ‘scanty’ information, and also ‘eminent Turkish women’, ‘those whose names have been preserved, but who were neither scholars nor officials’, and, finally, those mentioned in various sources that he had ‘not yet tackled’.[222] Including Uighurs, Qarluqs, Qanglis, Qipchaqs, Öngüts, Kereyids, and Naimans, the total for the entire Mongol/Yuan period is 646 (excluding Kereyids and Naimans, who may not, or may not all, have been Turkic, the total is 550).[223] This figure can be compared with one for the number of Huihui who were officials under the Yuan, compiled by Donald Leslie. He includes only those for whom biographical information is available, but draws upon quite numerous sources, both primary and secondary. His figure is 43.[224] Although the methods and sources used by de Rachewiltz and Leslie for the compilation of their figures undoubtedly differed, so that the numbers are not directly comparable, the large discrepancy between them is at least indicative. Once again, when it is taken into consideration that a number of Leslie’s Huihui were undoubtedly Turks, it is clear that Persians must have been greatly outnumbered by Turks in Yuan society.[225] It is perhaps also worthy of note that, of the 50 or so ‘eminent personalities of the early Mongol-Yuan period (1200–1300)’ whose biographies have been collected in In the Service of the Khan, only one (Sayyid Ajall Shams al-Din) might perhaps have any claim to have been Persian.[226]

I now come to the question of the inscriptions on paizi, and on standard weights. First of all, it must be said that they are very few in number. Although Liu Yingsheng has claimed that: ‘During the last 50 years many metal (bronze, silver or gold) paizi … (tablets of authority) were found with the above mentioned three languages [Mongolian, Chinese and Persian] written on them — some had even five languages’,[227] the fact is that the number of known extant Mongol paizi is very few. Moreover, only about four carry inscriptions in any language other than Mongolian, and only one is known with inscriptions, not in five languages, but in five scripts.[228] There is an important issue here, for it appears that scholars from China quite commonly confuse ‘language’ and ‘script’. For example, Liu Yingsheng refers to 'Phags-pa as a ‘language’, although it was no such thing.[229] It was a script, devised at the command of Qubilai Qa’an and named after the 'Phags-pa Lama. It was intended to be a universal script, that could be used to write any language, and was in fact often used for writing Mongolian, and sometimes also for writing Chinese, Sanskrit, Tibetan and Turkic.[230] Most extant Mongol paizi bear inscriptions in Mongolian in the 'Phags-pa script. An article published in 2003 lists, and gives brief descriptions of, seventeen extant Mongol paizi. This included almost all paizi known at the time worldwide.[231] In 2004, a monograph on inscriptions in Mongolian in the 'Phags-pa script described eleven paizi, and mentioned a further one.[232] The difference in number between these two publications is accounted for mainly by the fact that some paizi bear inscriptions only in Mongolian written with the Uighur script, and not with the 'Phags-pa script. In 2008, a list of eighteen paizi was published, but two were known only from illustrations in books, and one was fragmentary.[233] This list is not comprehensive, however. It does not include the paizi in the collection of the Metropolitan Museum of Art in New York, for example.[234] Additionally, there are two paizi in the collection of the History Museum of the Southwestern University in Yunnan.[235] There are, then, no more than about twenty Mongol paizi currently extant.[236]

The problem that scholars from China seem to have regarding ‘script’ and ‘language’, probably stems from the fact that modern Chinese does not make a clear distinction between the two. It is common, for example, to refer to the Chinese language as Zhongwen 中文, although, strictly speaking, this should mean ‘Chinese writing’. Another example of this confusion is that Huang Shijian says: ‘the so-called Tāzīk script … meant not the Arabic script, but Persian writing’.[237] Persian, of course, was written using Arabic script, so that this assertion makes no sense at all. This question affects the issue of these Persian inscriptions, for often, in Chinese, they are said to be in ‘Bosiwen 波斯文’, and whether this means ‘Persian language’ or ‘Persian script’ is uncertain.

The significance of these inscriptions is also unclear. They are very few, only about a dozen in total (mostly on weights, with a few on paizi), and very short, not more than five words each. A few short inscriptions in Persian may not mean very much at all. Most extant weights and paizi do not have any Persian on them. The use of Persian inscriptions, alongside Mongolian and Chinese, on standard weights and paizi may have been no more than an indication of the claim of the Great Qan to be supreme ruler of the entire Yeke Mongghol Ulus.[238] The Yuan emperor, after all, always claimed to be more than just the ruler of the Toluid Ulus in Mongolia and China. He was the Great Qan (Qa’an), supreme overlord of the whole Chinggisid Empire. Since Persian was undoubtedly the most important language of the Ilkhanate, part of this great empire, then its use on paizi and standard weights may have been simply a way of asserting this claim. The often symbolic nature of inscriptions of this kind is well exemplified by the inscriptions in Latin on current British coins. Latin is certainly not any kind of a lingua franca in the United Kingdom today. Closer to the period in question here, coins issued under Qara Khitai rule bore inscriptions in Chinese,[239] although there must have been very few Chinese speakers in the Qara Khitai empire. Even more relevant is the fact that coins with Chinese characters on them were struck in various parts of the Great Mongol Empire, well beyond the borders of China. Among these are silver-washed copper coins from Bukhara, in the Chaghatai Qanate, minted during the 1260s.[240] Also, two silver coins with inscriptions in Chinese characters, dating from about 1270, have been found at Khmelevka near Saratov, in the Jochid Ulus.[241] In the Ilkhanate, not only do Chinese characters occasionally appear on coins, but seals bearing inscriptions only in Chinese were used by the Ilkhans (they were very probably issued to them by the Great Qan in Dadu).[242] Yet no-one has ever suggested that Chinese was a lingua franca of the Yeke Mongghol Ulus. Such a suggestion would surely not be justified.

It must be noted, however, that the inscriptions on weights and paizi are not certainly Persian. It has also been suggested that they are, in fact, Turkic written in Chaghatai script, that is, the Perso-Arabic script adapted for writing Turkic. There has, indeed, been considerable controversy about this issue in Chinese publications. In their English-language publications, neither Huang Shijian nor Liu Yingsheng make any mention of this controversy. Huang no doubt felt that he had dealt with the question, in one of his Chinese-language publications, discussing the inscriptions on weights. He opines that the Perso-Arabic script had not been applied to writing Turkic before the fourteenth century, and that, in the conditions of the late 1200s and early 1300s, when the Great Qan was at war with Qaidu and the Chaghatai Qanate, it would be unlikely that the Chaghatai script would be used on weights produced in China.[243] However, as has already been noted above, Turkic was being written with the Perso-Arabic script at the court of Güyük Qa’an, in Mongolia, as early as 1246. The evidence for this, in the surviving letter from Güyük to the Pope, is incontrovertible. Indeed, the Perso-Arabic script began to be used for writing Turkic much earlier, as early as the eleventh century.[244] A document from Yarkand (in what is today Xinjiang), dated AH 515 [1121], is in Turkic written with the Arabic script.[245] Huang is therefore certainly wrong regarding dating. As it seems quite likely that the Perso-Arabic script was used by Muslim Turks throughout the Mongol empire, including Mongolia and probably also China, from at least as early as the 1240s onwards, his argument about the war with the Chaghatai Qanate is likewise unconvincing.

What is particularly striking is that the Perso-Arabic inscriptions on these weights were not only identified as Chaghatai, but were also read, and translated into Chinese.[246] How this could have been done, if the language had been wrongly identified, seems hard to comprehend. Yet Huang Shijian claims also to be able to read the inscriptions, as Persian.[247] This is certainly something of a mystery. Until this mystery is properly resolved, the inscriptions on these weights cannot be accepted as good evidence for anything. Interestingly, in his recent English-language paper about Persian in China, Liu Yingsheng makes absolutely no mention of the inscriptions on these weights.[248]

Liu mentions the inscriptions on paizi, however. As already stated above, he exaggerates the number, both of extant paizi and of Perso-Arabic inscriptions on them. It is perhaps worth repeating that, at least up to 2008, only four Perso-Arabic inscriptions on Mongol paizi had been reported. Liu also claims that:

As Persian had been the most important written language in Central Asia since the end of the 10th century and most of the Muslims in Yuan China came from Central Asia, they were strongly influenced by Iranian culture, and Persian became the common language among the Huihui population, and later even the mother tongue of many of their children and grandchildren before the process of sinification came to an end. Until the present day, there are still many Persian words and phrases in the daily Chinese language of the Huihui.[249]

He gives no evidence for these claims, which I consider to be overstated. In the absence of evidence, these assertions cannot be accepted.[250] It may well be that Persian was important as a written language of culture and learning in Central Asia, but it does not follow that it was also commonly spoken, as a lingua franca must be.[251] It is perhaps relevant to note here that, although there were schools where boys studied the Quran and other religious works, literacy among the Persian-speaking Tajiks of Central Asia in pre-modern times was very low, only something over two per cent.[252]

The paizi with inscriptions in five different scripts is a good example of the controversy surrounding these ‘tablets of authority’.[253] A very interesting article, published in a Chinese academic journal, reveals that five different Chinese experts had been asked, by the journal’s editors, to give their interpretations of the Perso-Arabic inscription on this paizi. All five gave different readings. One considered that the language of the inscription was Turkic, the other four took it to be Persian. One of these four, however, was able to read only one word of the inscription. There was also disagreement regarding the number of words in the inscription, two thinking that there were five, the others, only four. The editors of the journal expressed a certain degree of dismay at this: ‘Regarding these different readings, we of the editorial department now have no ability to discriminate between them, and at the moment also do not know where to find advice …’.[254] I personally do not have the linguistic competence to attempt to read the inscription, and I have no access to the original paizi (which I believe is now in a museum in Inner Mongolia, near where it was found). I can therefore offer no definite opinion about this issue. Liu Yingsheng’s comments on the inscription are of interest, however, and it is worth considering what he says. His conclusion is that the inscription is in Persian, but that there are errors in the writing of the letters, and that, although the vocabulary is Persian, it is not written in Persian word-order. He suggests that the word-order is that of Chinese. He further says that this must be because the paizi is from late in the Yuan period, presumably meaning that, during the later part of the dynasty, knowledge of correct Persian had deteriorated, and Chinese influence on the language had become strong.[255] His reading of the inscription has the merit of being very similar in meaning to the inscriptions in other languages on the same paizi. A sixth interpretation of this inscription was published a year later. Its two authors saw four words in the inscription, taking two of them to be different from all previous readings, and indicating some doubt regarding the correct reading of the second word. They considered the language of the inscription to be Persian, but did not address the issue of the non-Persian word-order. They took the last word to be a noun with the same meaning as paizi.[256] This did not, therefore, represent much of an advance on previously published opinions.

There are a number of points that seem worth making here. Firstly, Liu reads the last word of the inscription as ‘paiza’.[257] This is, of course, a loan-word from Chinese, and it occurs not only in Persian but also in Mongolian and Turkic.[258] This word is therefore ambiguous as regards determination of the language of the inscription. Liu sees a total of five words in the inscription, although all but one of the other experts considered that there were only four. Since there is no agreement about the reading of these words, and since there were certainly Persian loan-words in the Turkic of the period,[259] it seems very difficult to make any decision about the language of the inscription based only on vocabulary. What is perhaps more significant is what Liu says about the word-order. I strongly doubt that this results from Chinese influence. It is known that, during the Yuan period, Chinese was sometimes influenced by Mongolian syntax, presumably when official documents were being translated from Mongolian into Chinese,[260] but I am not aware of this kind of influence in the opposite direction. Indeed, since all important official documents were written in Mongolian, and then translated into other languages,[261] it would seem unlikely. The word-order of this inscription, if it is clearly not right for Persian, would, almost certainly, be the normal word-order of Turkic, with the principal noun, ‘paiza’ (or however it may be read), in final position, preceded by words qualifying it.[262] This obviously needs more research, but, at the moment, it seems to me that this inscription is at least as likely to be Turkic as Persian.[263] It would appear that, as with the inscriptions on the weights, the inscriptions on paizi are of very uncertain value as evidence for anything.

There is another issue regarding both the paizi and the weights. Some of them do not appear to have very good provenance, so that there is little assurance that they are all authentic. It is well known that a huge quantity of fake antiquities has been produced in China in the last few decades (and, indeed, much earlier).[264] Even dinosaur fossils have been faked, in considerable quantity.[265] Any object that does not have good provenance must, therefore, be open to a certain amount of suspicion. The provenance of some of the paizi, in particular, leaves considerable room for doubt. What is especially worrying is that, in Chinese publications, issues of provenance and authenticity are scarcely ever mentioned.[266] Indeed, it can be difficult to find precise information about the provenance of any particular item. The mere fact that about half of all known extant Mongol paizi have been found in China during the last thirty years is, in itself, worrying.

One especially fine, gold paizi is reported to have been discovered by a local farmer when digging sand for building from beside a river in the Qorchin Right Wing Front Banner in south-eastern Inner Mongolia in 1961.[267] He kept it, presumably secretly, for many years.[268] Eventually, his son sold it to a professor from Inner Mongolia University, in April 2000, no less than 39 years after its alleged discovery. In 1986, about 1.5 kilometres from a village in the Qorchin Right Wing Central Banner, a farmer found the five-script paizi discussed above.[269] Considering that this bronze paizi had presumably being lying in the ground for more than six centuries when it was discovered, it seems to be in very good condition. Of course, it is entirely possible that genuine objects could be found in such circumstances. Unfortunately, however, this kind of provenance provides very little assurance of authenticity.

It is also of significance that there are extant coins of the Mongol empire which bear inscriptions in Turkic written with Arabic script. A large issue of coins of the regency of Töregene, who was regent during the years 1241–46, bears the Turkic inscription, ‘Ulugh Mangyl Ulūs Bek’.[270] These coins were struck at mints in Transcaucasia and Iranian Azerbaijan, including Tabrīz, possibly at the behest of Baiju, the Mongol commander in that area.[271] Coins struck in Samarkand during the 1220s often bear Persian inscriptions, but about three decades later similar inscriptions are in Turkic.[272] Mongol coins frequently carry inscriptions in Mongolian in Uighur script, and commonly also in Arabic (language and script). Coins of the Yuan empire often bear inscriptions in Mongolian written with the 'Phags-pa script, which occasionally appears on coins of other Qanates. Chinese characters are also of common occurrence on Yuan coins. Persian inscriptions appear sporadically, but even coins from the Ilkhanate normally bear inscriptions in Mongolian and Arabic, and only sometimes in Persian.[273] This numismatic evidence does not suggest that Persian was used widely outside the Ilkhanate during the Mongol period. It was very probably only during the Timurid period and later that Persian became commonly used in Central Asia: in post-Mongol Iran and Central Asia, ‘[t]he middle of the 15th century can be considered the turning point in the struggle of Persian for leadership in coin design’.[274] Inscriptions on coins do not necessarily bear much relation to the language(s) actually spoken in the regions where the coins circulated, however. Persian verses appeared on the coins of the Muslim rulers of India, ‘for whom (as for the majority of their subjects) Persian was not even a native language’.[275]

I come now to the final piece of evidence which, according to Huang Shijian, proves the importance of Persian in the Yuan empire, and that is, the Huihui Guozi Xue. Liu translates this ‘School of Persian Language’.[276] Such a translation really cannot be justified. A much more accurate translation is ‘Muslim National College’,[277] and there is, in fact, absolutely no evidence at all that it was involved with the teaching of Persian. I have looked at all the references that I have been able to trace to this Muslim National College, in the Yuan shi[278] and other sources,[279] and I have found no mention of what languages were studied in it. The sources say that it was set up for the study and teaching of the ‘Istifi’ (Yisitifei 亦思替非, 伊斯提费) script.[280] There has been quite a lot of discussion of what exactly this ‘Istifi’ script was, with some rather bizarre theories advanced,[281] but there should be no real difficulty with this expression. ‘Istifi’ is an Arabic word. ‘Istifa … comes from the root verb [in Arabic] safa, which means to be clear or pure, or to select the best. In the Qur’an, Allah istifa (chose) his messengers and prophets … ‘.[282] Thus, the ‘Istifi’ script was the script that had been ‘chosen’ to write down the words of God in the holy Quran; that is, the Arabic script.[283]

Since there is no indication in the sources of what language or languages were written with this ‘Istifi’ script, there is really no point in speculating on the subject. Persian may well have been among them, but so may Arabic and also Turkic. It must be said that the Muslim National College does not seem to have been very important. According to Liu Yingsheng: ‘about 50 persons of this agency were staff and government-sponsored students; meanwhile dozens were students who paid their own expenses’.[284] However, this is not exactly what the Yuan shi actually says:

In the second year of the Taiding 泰定 reign-period [1325], in spring, in the intercalary first month, because this year the sons and younger brothers of the nobility and the sons of ordinary people who entered the College to study were numerous, the teachers and students were more than fifty. Apart from twenty-seven who already were supplied with food and drink [presumably this means that they were given a subsistence allowance by the government], there were a teaching assistant and twenty-four students requiring official support, and it was ordered to supply them.[285]

Several things are clear from this. Firstly, it was exceptional for there to be as many as more than 50 students and teachers in the College. Normally, the number would have been less, and probably significantly less, otherwise it would not have been noteworthy for there to be more than 50. Secondly, there were no ‘students who paid their own expenses’. There were only those who did not yet have official support, but were granted it. Moreover, they were included in the figure of more than 50, not additional to it. Thus, it can be seen that normally, there were probably only some three or four dozen students and teachers in the Muslim National College. This may be compared with the situation in the Mongolian National College, which ‘by 1315 had places for 100 students, although enrollment sometimes ran to 200 or 300’.[286]

The information translated above about the students at the Muslim National College can be analyzed further. Those students already receiving government support must have been continuing students, who had entered the College before 1325. The 24 students requiring similar support were the new intake in 1325. Perhaps there was some question regarding whether the government would pay to support all of them, as they were more numerous than usual. The 27 continuing students must have been from more than one year’s intake, otherwise 24 would not have been an exceptionally large enrolment. Thus, in a normal year, probably about thirteen or fourteen new students entered the College. If they usually each spent three years in the College, then the normal number of students would have been about 40. With perhaps as many as half a dozen teaching staff, this would give a normal complement for the College of about 45, roughly ten or a dozen less than in 1325.

The Yuan shi also states that the students from the College filled the posts of translators (yishi 譯史),[287] as needed, in the various offices of government.[288] The work of these translators would mainly have consisted in translating official documents. Perhaps, therefore, the Muslim National College was indeed solely concerned with teaching Arabic script. The students who entered the College may have already spoken various languages. What they needed was training in how to write them well. Perhaps some of them knew how to write Turkic in the Uighur script, but wanted to learn to write it with Arabic script, too. This is largely speculation, of course, but the point is that absolutely nothing is said in the sources about the Muslim National College being involved with anything other than script. The claim made by Hyunhee Park, citing Liu Yingsheng as source, that: ‘Scholars attending these Muslim institutes at the Mongol court, according to Chinese sources, conducted their discourse in the ‘Huihui’ or ‘Muslim’ language, which in most cases meant Persian’,[289] is untrue, and is not, in fact, supported by Liu’s statements. No Chinese sources say anything of this kind.

Morgan, following Huang Shijian, says that the Muslim National College was ‘renamed a Directorate of Education in 1314’.[290] This is not exactly correct. In fact, a Muslim National Institute (Huihui Guozi Jian 回回國子監)[291] was created in 1314, which was not merely the National College under a new name, but an additional, higher level, body. What Huang and Morgan failed to note, however, is that this National Institute had a very short existence, as it was abolished in 1320.[292] Liu says that: ‘Emperor Renzong 仁宗 in 1314 ordered the re-establishment of an official post of ‘Supervisor of the Persian School’ (Huihui guozi jianguan 回回國子監官) to supervise the Persian language education in the school’.[293] The term Huihui guozi jianguan does not mean ‘Supervisor of the Persian School’, however (here it is probably plural, and means ‘officials of the Muslim National Institute’), and the supposed ‘re-establishment’ is an error, based on a misunderstanding of the original Chinese text.[294]

In the conclusion of his paper, Professor Morgan brings up a few new pieces of evidence for the importance of Persian in the Yuan empire. He points out that, in a letter written in 1306, John of Monte Corvino says: ‘I have had six pictures made of the Old and New Testaments for the instruction of the ignorant, and they have inscriptions in Latin, Turkish and Persian …’.[295] Again, this proves very little. Latin certainly was not a widely used language in the Yuan empire, and Turkish appears here alongside Persian. It seems likely that Persian was included because it was a language of the Nestorian Christians,[296] whom John no doubt hoped to attract to the Latin Church. Giving Rossabi as his source, Morgan also asserts that even “so quintessentially Chinese a dish as jiaozi” may have been of Persian origin, and have entered China during the Yuan dynasty. This is almost certainly untrue, for dumplings (jiaozi 餃子, 角子)[297] seem to have existed in China before the Mongol invasions. A work of the early Song period records that: ‘during the reign of Ren Zong 仁宗 (1023–63), when the spring examination was held for the Jin Shi 進士 degree’ the ladies of the court assembled to watch. The Jin Shi examinees were presented with cake dumplings (bing jiaozi 餅角子), and the examining officials were given ‘seven-treasures tea’ (qi bao cha 七寶茶).[298] This does not, of course, completely rule out a Persian origin for jiaozi, but it does make it very unlikely that they were introduced during the Yuan dynasty.

A check of Morgan’s source, however, reveals that Rossabi does not, in fact, state that jiaozi may have come from Persia. His actual words are ‘jiaozi (dumplings) may have come to China from West Asia during the Mongol era’.[299] A further check, back to Rossabi’s source, reveals that it suggests that jiaozi were ‘probably inspired’ by ‘Arab cooking’.[300] So, jiaozi are not from Persia, after all. Rossabi’s source was Buell and Anderson’s translation of, and commentary on, the Yinshan zhengyao 飮膳正要. An examination of this work shows that it provides very little support for the thesis that Persian was an important lingua franca in the Yuan empire. On the contrary, it indicates the predominance of Turkic. Buell and Anderson list and count the words of foreign origin that occur in the Yinshan zhengyao, and they are overwhelmingly Turkic: ‘Turkic terminology … dominates with 36 out of 49 words’.[301]

I am not trying, in this paper, to deny that Persian had any place at all in the Yuan empire. Clearly, there were Persians in China during the Yuan period, and the Persian language was of some importance as a language of learning. It was important in astronomy, for example, and to some extent also in medicine.[302] There were periods when Islam, and, quite probably, therefore, also Persians and the Persian language, were particularly influential. For example, during the reign of the Taiding emperor 泰定皇帝, Yesün Temür (1323–28), Muslims were very influential in the government. A Muslim of uncertain origins, Daula Shah, ‘was apparently the moving spirit behind the administration’.[303] On the other hand, there were also times when Islam was discouraged.

Although there were Muslims in the entourage of Chinggis Qan from an early date, and even among those who ‘drank the water of Baljuna’ with him,[304] at one time he prohibited the slaughter of animals in the Muslim fashion.[305] The Mongols were clearly often uncomfortable with Islam, because many of its tenets conflicted with their own customs and beliefs. As they were the conquerors, they saw no reason to tolerate what they sometimes perceived to be insulting behaviour on the part of those they had conquered:

Among all the [subject] alien peoples only the Hui-hui say ‘we do not eat Mongol food’. [Činggis Qa’an replied]: ‘By the aid of Heaven we have pacified you; you are our slaves. Yet you do not eat our food or drink. How can this be right?’ He thereupon made them eat. ‘If you slaughter sheep, you will be considered guilty of a crime.’ He issued a regulation to that effect. … [In 1279/1280 under Qubilai] all the Muslims say: ‘If someone else slaughters [the animal], we do not eat.’ Because the poor people are upset by this, from now on, Mussulman Hui-hui and Chu-hu [Zhuhu] 术忽 (Jewish) Hui-hui, no matter who kills [the animal] will eat [it] and must cease slaughtering sheep themselves and must cease the rite of circumcision.[306]

These harsh prohibitions may often not have been strictly enforced, or not enforced for long, but they clearly show the difficulty that the Mongols (and the Chinese, who were probably among ‘the poor people’) had with accepting Islamic practices. It would seem that Qubilai Qa’an may have felt a particular dislike for them. In 1290, a senior Muslim official of the Jiang-Huai 江淮 Branch Secretariat reported that officials of the public granaries had stolen grain and embezzled money, and requested that they be punished ‘according to Song law’, by carving their faces[307] and cutting off their hands. Qubilai replied: ‘This is Islamic law!’ and would not permit it.[308]

It is also recorded that on two occasions, in 1292 and 1293, Muslim merchants offered large pearls to the Qa’an, but Qubilai refused them, saying that pearls were a waste of money, which could be better spent helping the poor.[309] Whether it was for a similar reason, or perhaps because of the war with Qaidu and the Chaghatai Qanate, in 1281, Qubilai forbade all Muslims of the north-western border area to cross the border to trade. Ten years later, in 1291, he forbade Mongols to travel to Muslim regions as merchants.[310] It seems that Qubilai was prepared to restrict trade if he felt it to be necessary, and did not always look favourably on the activities of Muslim merchants.

Although Islam prospered in the other Mongol Qanates, and, sooner or later, the rulers of the Ilkhanate, the Jochid Ulus and the Chaghatai Qanate all became Muslims, in the Yuan empire it was Tibetan Buddhism that won most favour with the Qa’ans. The 'Phags-pa Lama, who had been greatly esteemed by Qubilai Qa’an, posthumously became a figure of veneration. In the first year of the reign of Ying Zong 英宗 (1320), ‘it was commanded that a Hall of the Imperial Preceptor 'Phags-pa should be built in every prefecture. Its style should be similar to that of the Confucian Temples, but grander’.[311] In at least one instance, this command redounded badly for Islam, for in the following year ‘the mosque in Shangdu was demolished, and its site was used for a Hall of the Imperial Preceptor’.[312]

There were also general reductions in the privileges of the Muslim community during the Yuan period. Whereas at first Muslims had enjoyed exemptions from taxation and from corvée labour, and had been allowed a considerable degree of self-governance, under their qadis, these privileges were gradually abolished after 1310, and particularly during the 1320s. The death of Yesün Temür in 1328 was followed by the execution of Daula Shah and at least one other high-ranking Muslim official.[313] Muslims and Islam were often unpopular with the masses in China, too. One of the dramas of the time contains a threat by one character against another: ‘I’ll sell you to a Muslim, a Tartar, or a Jurchen!’[314] Clearly, being sold to a Muslim was not thought to be a pleasant fate. There is also the well-known note by Tao Zongyi on the disaster which befell a Muslim wedding party in the Ba Jian Lou 八間樓 in Hangzhou. So many curious onlookers climbed onto the roof that the building collapsed under their weight, killing the bride, groom and many others. Tao quotes a thoroughly unpleasant satirical poem about the incident, which displays obvious prejudice against Muslims.[315]

It seems to me that several conclusions are inevitable. Firstly, Muslims were not the majority of the Semu ren in the Yuan empire. The largest single group of Semu ren was undoubtedly the Turks. Indeed, Turks were a major element in the entire Yeke Mongghol Ulus. Golden quotes al-‘Umarī regarding the Mongols and the Qipchaqs:

this country was formerly the land of the Qibjâq. When the Tatârs inun-dated it, the Qibjâq became their subjects. Then, they mixed with them and intermarried with them. The land was victorious over natural disposition (jibillah) and origins. All (of them) became like the Qibjâq, as if they were of one stock (jins wâhid), because the Mogul lived in the land of the Qibjâq and (because) of their marital ties with them and their community in their land.[316]

This process, or a very similar one, affected two of the major parts of the Mongol empire, the Chaghatai Qanate and the Jochid Ulus. The Mongol conquests, indeed, added to an influx of Turks into Central and Western Asia, which had begun centuries before the time of Chinggis Qan.[317] Various dialects of Turkic, most of them mutually intelligible, became the common language of a wide belt of country, from the Tarim Basin all the way to Anatolia and the northern shores of the Black Sea. Cumans (Qipchaqs) settled in Hungary and the Balkans,[318] so that Turkic was a language even of central Europe. Thus, all the way from Europe to China, there were Turks, and speakers of Turkic.

It is also quite likely that Turks were the majority of the Muslims in the Yuan empire. This cannot be asserted with complete assurance, but it is clear that at least a substantial proportion of Muslims in the Far East were Turks. Since it is also clear that many of the non-Muslim Semu ren were Turks, it is obvious that Turks greatly outnumbered Persians in the Yuan empire. Persian was an important language in one part of the Yeke Mongghol Ulus, the Ilkhanate, and was likely the lingua franca of the maritime trade routes from the Persian Gulf to the south-east coast of China. It was not a major language elsewhere, however. In the Jochid Ulus and the Chaghatai Qanate, Turkic was the predominant language. In the Yuan empire, Turkic was the predominant language of the Semu ren. There is absolutely no good evidence that Persian was an ‘official’ language of the Yuan court. This claim is based on poor evidence, all of which has been shown in this paper to be either invalid or, at best, of very dubious value.

There is a further important point to be made here. For far too long, the study of the Mongols, their conquests, and their empire has been dominated by scholars of Persian. It has been claimed that the Persian sources, particularly Rashīd al-Dīn’s Jāmi’ al-Tavārīkh, are the most important for this study.[319] Of course, the works of Juvainī, of Rashīd al-Dīn, of Wassāf, and of one or two other Persian historians are extremely important. The fundamental sources, however, are undoubtedly Chinese. This should surprise no one. After all, the Mongols lived in close proximity to the Chinese from a very early date. Indeed, Chinggis Qan himself was at one time a vassal of the Jin empire of northern China.[320] The Chinese also kept extremely good historical records, and had been doing so continuously for centuries before the rise of the Mongols. Long before anyone in Persia had even heard of the Mongols, the Chinese were writing about them.[321] While it is true that the Yuan shi, the official dynastic history of the Mongol empire in China, is deficient in various ways, it is still a voluminous document containing a vast amount of information.[322] More to the point, it is by no means the only Chinese source for the period. This subject requires a separate paper, which may well need to be even longer than this one. Suffice it to say, the status of Persian in the Mongol empire has been greatly exaggerated, and so has the importance of Persian sources for Mongol studies.[323] As Rashīd al-Dīn himself wrote: ‘I realized that they [the Chinese] are masters in all fields of knowledge’.[324]

Stephen G. Haw
Independent scholar



1 D.O. Morgan, ‘Persian as a Lingua Franca in the Mongol Empire,’ in eds B. Spooner and W.L. Hanaway, Literacy in the Persianate World: Writing and the Social Order (Philadelphia: University of Pennsylvania Press, 2012), pp.160–70.

2 P.B. Golden, An Introduction to the History of the Turkic Peoples (Wiesbaden: Harrassowitz, 1992), p.195.

3 The original Lingua Franca was a language used in the eastern and southern Mediterranean region as a medium of principally oral communication, from the time of the Crusades until the early twentieth century. Based mainly on the Genoese dialect of Italian, it included vocabulary from other Romance languages, Arabic, Greek, and Turkish; see European Union, Directorate-General for Translation, Lingua Franca: Chimera or Reality? (Luxembourg: Publications Office of the European Union, 2011), pp.19–20. I use the term lingua franca in this article in the sense of a language widely used, by both native and non-native speakers, as a common language for mainly oral communication.

4 Morgan, ‘Persian as a Lingua Franca,’ p.161.

5 Only a very small fraction of the Chinese sources for the Mongol/Yuan period have ever been translated into any other language.

6 H. Yule, The Book of Ser Marco Polo, the Venetian, Concerning the Kingdoms and Marvels of the East (London: John Murray, 1871), Vol.1, p.cxxxv; H. Yule, The Book of Ser Marco Polo, the Venetian, Concerning the Kingdoms and Marvels of the East, 3rd ed. (rev. H. Cordier) (London: John Murray, 1903), Vol.1, p.110.

7 H. Yule, Marco Polo,Vol.1, p.401n; H. Yule, Marco Polo, 3rd ed. (rev. H. Cordier), Vol.1, p.448n.

8 It should be noted that, at the time, China was generally very little known and studied in Europe. It was not until the 1870s, for example, that a chair of Chinese was established at the University of Oxford. There had been one at King’s College, London, since 1847; see L. Ride, ‘Biographical Note,’ in J. Legge, The Chinese Classics, Vol.1, Confucian Analects, The Great Learning, The Doctrine of the Mean (Hong Kong: University Press, 1960), p.18. France led the way in academic studies of China, with a Chair of Chinese at the Collège de France from 1814; see A.-L. Dyck, ‘La Chine hors de la philosophie: essai de généalogie à partir des traditions sinologique et philosophique françaises au XIXe siècle,’ Extrême-Orient, Extrême-Occident 27 (2005): 13–47, at p.28. In the mid-nineteenth century, Chinese studies were still at an early stage in Europe.

9 F.W. Cleaves, ‘The Biography of Bayan of the Bārin in the Yüan Shih,’ Harvard Journal of Asiatic Studies 19 (1956): 185–303, at p.187n.

10 G. Pauthier, Le Livre de Marco Polo, citoyen de Venise, conseiller privé et commissaire impériale de Khoubilaï-Khaân; (Paris: Firmin Didot, 1865), Vol.1, p.23nn.

11 P. Pelliot, ‘Kao-Tch’ang, Qoco, Houo-Tcheou et Qarâ-Khodja,’ Journal Asiatique, 10.19 (1912): 579–603, at p.592.

12 H. Cordier, Ser Marco Polo: Notes and Addenda to Sir Henry Yule’s Edition, Containing the Results of Recent Research and Discovery (London: John Murray, 1920), p.74.

13 P. Pelliot, Notes on Marco Polo (Paris: Imprimerie Nationale, 1959–73), Vol.1, p.424; see also ibid., Vol.1, pp.96, 170, 366; Vol.2, p.813.

14 F. Wood, Did Marco Polo Go to China? (London: Secker & Warburg, 1995), p.61.

15 I have already made this argument; see S.G. Haw, Marco Polo’s China: A Venetian in the Realm of Khubilai Khan (London and New York: Routledge, 2006), p.62.

16 Haw, Marco Polo’s China, p.61.

17 A.C. Moule and P. Pelliot, Marco Polo: The Description of the World (London: Routledge, 1938), Vol.1, p.84; H. Yule, Marco Polo, 3rd ed., Vol.1, p.25.

18 Wood, Did Marco Polo Go to China?, p.42; J. Larner, Marco Polo and the Discovery of the World (New Haven: Yale University Press, 1999), p.54.

19 Pelliot, Notes, Vol.1, p.256. Pelliot sees no difficulty in deriving ‘Ciandu’ directly from Chinese; it may be noted in passing that Marco’s use of a final –n in the first syllable, rather than –ng, is a normal feature of Romance languages, which generally lack final –ng; the French still usually refer to Beijing as ‘Pékin’, for example, while modern Italian usage is ‘Pechino’.

20 Song Lian 宋濂 et al. eds., Yuan shi 元史 (hereafter YS) (Beijing: Zhonghua shuju, 1976), Vol.1, juan (hereafter j.) 5, p.92 ; Vol.5, j.58, pp.1349–50; Liu Yingli 劉應李, Da Y uan hunyi fangyu shenglan 大元混一方輿勝覽, (rev. Zhan Youliang 詹友諒) (Chengdu: Sichuan daxue chubanshe, 2003),Vol.1, p.41.

21 Wang Yun 王惲, Qiujian xiansheng daquan w enji 秋澗先生大全文集, j.1, p.10b, in Si bu congkan chu b ian 四部叢刊初編 [facsimile of Ming edition] (Shanghai: Shangwu yinshuguan, 1929), refers to Shangdu in the second year of the Zhongtong 中統 reign-period, that is, 1261.

22 D.M. Farquhar, The Government of China Under Mongolian Rule: A Reference Guide (Stuttgart: Steiner, 1990), p.427; F.W. Cleaves, ‘The Sino-Mongolian Inscription of 1362 in Memory of Prince Hindu,’ Harvard Journal of Asiatic Studies 12(1949): 1–133, at p.83.

23 Rashīd al-Dīn (trans. W.M. Thackston), Rashiduddin Fazlullah’s Jami’u’t-Tawarikh: Compendium of Chronicles (Cambridge: Harvard University, Dept. of Near Eastern Languages and Civilizations, 1998–99),Vol.1, p.220; Vol.2, p.374; Rashīd al-Dīn (trans. J.A. Boyle), The Successors of Genghis Khan (New York: Columbia University Press, 1971), p.146.

24 Rashīd al-Dīn (trans. Thackston), Vol.2, p.384; Boyle, Successors, p.164 and note.

25 Pelliot, Notes, Vol.2, p.803.

26 Haw, Marco Polo’s China, p.97.

27 Pelliot, Notes, Vol.2, p.814.

28 A. Schuessler, ABC Etymological Dictionary of Old Chinese (Honolulu: University of Hawai‘i Press, 2007), pp.528, 558.

29 Boyle, Successors, p.283.

30 Rashīd al-Dīn (trans. Thackston), Vol.2, p.446.

31 G. Ferrand, Voyage du Marchand Arabe Sulayman en Inde et en Chine, R édigé en 851, Suivi de Remarques par Abu Zayd Hasan (vers 916) (Paris: Bossard, 1922), pp.77, 86, 92, 105.

32 E. de la Vaissière, Sogdian Traders: A History (Leiden: Brill, 2005), p.22.

33 YS, Vol.1, j.7, p.143.

34 YS, Vol.2, j.14, p.302.

35 Boyle, Successors, pp.323–24; Rashīd al-Dīn (trans. Thackston), Vol.2, p.465.

36 Pelliot, Notes, Vol.1, pp.603–04.

37 J.F. Rock, The Ancient Na-Khi Kingdom of South-west China (Cambridge: Harvard University Press, 1947), Vol.1, p.42. I have personally seen this in southern Yunnan during the 1980s.

38 Moule and Pelliot, Marco Polo, Vol.1, p.287; Yule, Marco Polo, 3rd ed., Vol.2, p.101.

39 Sayyid Ajall’s biography, followed by that of his son, is in YS, Vol.10, j.125, pp.3063–67; see also I. de Rachewiltz, et al. ed., In the Service of the Khan (Wiesbaden: Harrassowitz, 1993), pp.466–79.

40 Pelliot, Notes, Vol.2, p.812.

41 Ibid., pp.652–61; more recently, Sanping Chen has argued that this term was also ‘widely used historically in various nomadic regions bordering the Chinese heartland’; see S. Chen, Multicultural China in the Early Middle Ages (Philadelphia: University of Pennsylvania Press, 2012), pp.124–34.

42 Morgan, ‘Persian as a Lingua Franca,’ pp.161–62.

43 Haw, Marco Polo’s China, p.61.

44 Morgan, ‘Persian as a Lingua Franca,’ p.169n.

45 Ibid., p.162.

46 Or, more accurately, any arthropod, worm, or similar creature.

47 He Guangyuan 何光遠, Jian jie lu 鑒誡錄, in Zhibuzu zhai congshu 知不足齋叢書 (Beijing: Zhonghua shuju, 1999), Vol.8, j.6, p.35

48 Zhao Yuyu 趙與裕, Xinsi qi Qi lu 辛巳泣蘄録 in Si ku quan shu cun mu congshu, s hi bu 四庫全書存目叢書, 史部, [facsimile of Ming MS] (Ji’nan: Qilu shushe, 1997), Vol.45, p.79.

49 On Nangiyas, or Nangɣiyas, see A. Mostaert and F.W. Cleaves, Les lettres de 1289 et 1305 des ilkhan Arɣun et Ölǰeit ü à Philip pe le Bel (Cambridge: Harvard University Press, 1962), p.73; and also P. Pelliot, ‘Sur quelques mots d’Asie centrale attesté dans les textes Chinois,’ Journal Asiatique, 11.1 (1913): 451–69, at pp.460–66.

50 For example, Rashīd al-Dīn (trans. Thackston), Vol.2, pp.413, 415, 438–39; Boyle, Successors, pp.223–24, 270–71, 273.

51 For example, Rashīd al-Dīn (trans. Thackston), Vol.2, pp.396, 440, 441; Boyle, Successors, pp.189, 273, 275.

52 Rashīd al-Dīn (trans. Thackston), Vol.1, p.154. This clearly shows that Rashīd al-Dīn himself considered ‘Manzi’ to be a Chinese term.

53 ‘Ala-ad-Din ‘Ata-Malik Juvainī (trans. J.A. Boyle), The History of the World-Conqueror (Cambridge: Harvard University Press, 1958), Vol.1, p.256; Vol.2, p,596.

54 Ibid., Vol.1, p.201,Vol.2, p.607.

55 Minhaj-ud-Din Juzjani (trans. H.G. Rav erty), Tabakat-i-Nasiri: A General History of the Muhammadan Dynasties of Asia … (London: Asiatic Society of Bengal, 1881), Vol.1, p.383.

56 Hamd-Allah Mustawfi, The Geographical Part of the Nuzhat al-Qulub, trans. G. Le Strange (Leiden: Brill, and London: Luzac, 1919), pp.10, 254.

57 For example, Moule and Pelliot, Marco Polo, Vol.1, pp.264, 265.

58 W.S. Coblin, A Handbook of 'Phags-pa Chinese (Honolulu: University of Hawai‘i Press, 2007), pp.124, 143.

59 Haw, Marco Polo’s China, p.61.

60 H. Yule, Cathay and the Way Thither, rev. H. Cordier (London: Hakluyt Society, 1915), Vol.2, p.115.

61 Haw, Marco Polo’s China, pp.1–2.

62 Moule and Pelliot, Marco Polo, Vol.1, p.227; Yule, Marco Polo, 3rd ed., Vol.1, p.397.

63 J.L. Schrader, ‘A Medieval Bestiary,’ The Metropolitan Museum of Art Bulletin, XLIV(1) (1986), p.3.

64 This illustration can be viewed online at: < documentation/enlumine/fr/BM/douai_ 552-01.htm>, viewed 17 Dec. 2012.

65 The relevant page of the manuscript can be viewed online at: < manuscripts/Viewer.aspx?ref=royal_ms_ 12_c_xix_f028r>, viewed 17 Dec. 2012.

66 Further information about tigers in mediaeval bestiaries is available online at: <>, viewed 17 Dec. 2012; see also F. McCulloch, Mediaeval Latin and French Bestiaries, (Chapel Hill: University of North Carolina Press, 1962), pp.176–77.

67 J.R. Allen, Early Christian Symbolism in Great Britain and Ireland Before the Thirteenth Century (London: Whiting, 1887), p.357.

68 Haw, Marco Polo’s China, pp.60–63, 96–97.

69 Moule and Pelliot, Marco Polo, Vol.1, p.95.

70 T.F. Carter, The Invention of Printing in China and its Spread Westward, 2nd ed. (rev. L.C. Goodrich) (New York: Ronald Press, 1955), pp.140–48.

71 Moule and Pelliot, Marco Polo, Vol.1, p.257; Pelliot, Notes, Vol.1, pp.8–9.

72 Rashīd al-Dīn (trans. Thackston), Vol.2, p.384; Boyle, Successors, p.165 and note.

73 M. Rossabi, Khubilai Khan: His Life and Times (Berkeley: University of California Press, 1988), pp.13–14; YS, Vol.1, j.2, p.35.

74 I. de Rachewiltz, ‘Turks in China Under the Mongols: A Preliminary Investigation of Turco-Mongol Relations in the 13th and 14th Centuries,’ in ed. M. Rossabi, China Among Equals: The Middle Kingdom and its Neighbors, 10th–14th Centuries (Berkeley: University of California Press, 1983), p.301; YS, Vol.10, j.125, p.3070; on the role of Uighurs in Zhending, see also M.C. Brose, Subjects and Masters: Uyghurs in the Mongol Empire (Bellingham: Center for East Asian Studies, West Washington University, 2007), pp.93–94, 96–105.

75 de Rachewiltz, ‘Turks in China under the Mongols,’ p.301; YS, Vol.10, j.124, p.3059.

76 de Rachewiltz, ‘Turks in China under the Mongols,’ p.286; YS, Vol.11, j 134, p.3246.

77 Moule and Pelliot, Marco Polo, Vol.1, p.265.

78 Pelliot, Notes, Vol.1, p.7.

79 Ibid., Vol.1, pp.140–41.

80 Moule and Pelliot, Marco Polo, Vol.1, p.212. The Chinese name was Dadu or Daidu 大都, the Mongolian version of this was Taidu.

81 Pelliot, Notes, Vol.2, p.843.

82 Schuessler, Dictionary of Old Chinese, p.202.

83 de Rachewiltz, ‘Turks in China under the Mongols,’ p.287; Golden, History of the Turkic Peoples, p.285; C.V. Findley, The Turks in World History (Oxford: University Press, 2005), p.87.

84 Brose, Subjects and Masters, p.77, says: ‘the Uyghurs became arguably the single most important sub-group of the Semuren administrators …’.

85 Ibid., pp.89n, 98, 112; Brose’s book about the Uighurs in the Yuan empire provides many examples of the importance of this Turkic people to the Mongols. It is unlikely that a similar book could be written about Persians in the Yuan empire.

86 Moule and Pelliot, Marco Polo, Vol.1, pp.177–78.

87 Pelliot, Notes, Vol.1, p.65.

88 G. Clauson, An Etymological Dictionary of Pre-Thirteenth-Century Turkish (Oxford: University Press, 1972), p.319; Clauson’s ‘Pterocles ahenarius’ is an error for Pterocles arenarius, now usually considered a subspecies of Pterocles orientalis, the Black-bellied (or Large) Sandgrouse.

89 Haw, Marco Polo’s China, p.129.

90 Moule and Pelliot, Marco Polo, Vol.1, p.230.

91 Pelliot, Notes, Vol.2, pp.859–60.

92 Moule and Pelliot, Marco Polo, Vol.1, p.182.

93 Pelliot, Notes, Vol.1, p.48.

94 Clauson, Pre-Thirteenth-Century Turkish, p.216.

95 Pelliot, Notes, Vol.1, pp.50-51.

96 Moule and Pelliot, Marco Polo, Vol.1, p.310.

97 Farquhar, Government of China, p.170.

98 There is a biography of Bayan in de Rachewiltz, et al. eds, In the Service, pp.584–607; see also F.W. Cleaves, ‘The biography of Bayan,’ pp.185–303.

99 Moule and Pelliot, Marco Polo, Vol.1, p.303.

100 Pelliot, Notes, Vol.2, pp.825–26.

101 Su Tianjue 蘇天爵, ed., Guochao wenlei 國朝文類, in Si bu congkan chu b ian 四部叢刊初編 [facsimile of Yuan ed.] (Shanghai: Shangwu yinshuguan, 1929).

102 Moule and Pelliot, Marco Polo, Vol.1, p.358.

103 Farquhar, Government of China, p.171; Pelliot, Notes, Vol.2, p.871.

104 YS, Vol.1, j.10, p.198.

105 Moule and Pelliot, Marco Polo, Vol.1, pp.216, 221.

106 Pelliot, Notes, Vol.2, p.815; Pelliot may be wrong in suggesting an original ‘quesictan’ in Marco’s book, as the final ‘-g’ of kešig was very possibly more or less silent. It does not appear in the Chinese transcription of kešig, qiexue 怯薛; the second character never ended with –g or –k, but in early times had a final –t, according to B. Karlgren, Grammata Serica Recensa (Stockholm: Museum of Far Eastern Antiquities, 1964), pp.88–89.

107 Moule and Pelliot, Marco Polo, Vol.1, p.179.

108 Pelliot, Notes, Vol.2, p.742.

109 Moule and Pelliot, Marco Polo, Vol.1, p.201.

110 Pelliot, Notes, Vol.2, p.823

111 Moule and Pelliot, Marco Polo, Vol.1, pp.163ff, 192ff; et passim.

112 Pelliot, Notes, Vol.1, pp.302–03; C.P. Atwood, Encyclopedia of Mongolia and the Mongol Empire (New York: Facts on File, 2004), pp.302–03. A. Vovin, ‘Once Again on the Etymology of the Title Qaɣan,’ Studia etymologica cracoviensia XII (2007): 177–87, at pp.180–84, has suggested that these words originated from a Yeniseian language.

113 See Haw, Marco Polo’s China, pp.96–97. It must be noted that the pronunciations of the two Jin was probably less similar in the thirteenth century than today, but dialect pronunciations may well have been closely similar. It is of course normal for the written and spoken forms of a language to differ, and it is not entirely clear exactly how the 'Phags-pa transcriptions of Chinese relate to the contemporary spoken language; it is possible that they represent a pronunciation that was already archaic in the thirteenth century. See Coblin’s remarks in Chapter IV of his Handbook of 'Phags-pa Chinese, especially pp.72–74.

114 Morgan, ‘Persian as a Lingua Franca,’ p.164.

115 B.G. Fragner, ‘Shah Ismail’s Fermans and Sanads: Tradition and Reform in Persophone Administration and Chancellery Affairs,’ Journal of Azerbaijani Studies I (1998): 35–47, at p.35. Fragner asserts that Persian replaced Arabic as the principal language of government only under the Ilkhans, the earliest known royal ferm ā n in the Persian language dating from 1233.

116 On the failure to distinguish differences among Muslims, see M. Rossabi, ‘The Muslims in the Early Yuan Dynasty,’ in ed. J.D. Langlois, China Under Mongol Rule (Princeton: University Press, 1981), pp.259–60. On the various terms used during the Yuan period for Muslims and others from the Western Regions, and their meanings, see D.D. Leslie, Islam in Traditional China (Canberra: College of Advanced Education, 1986), pp.195–96.

117 C. Melville, ‘The Chinese-Uighur Animal Calendar in Persian Historiography of the Mongol Period,’ Iran: Journal of the British Institute of Persian Studies 32 (1994): 83–98, at p.95.

118 This was the title of the Uighur ruler.

119 Ruler of the Qarluq Turks.

120 YS, Vol.1, j.7, p.128; T.T. Allsen, ‘The Yuan Dynasty and the Uighurs of Turfan in the 13th Century,’ in ed. Rossabi, China A mong Equals, p.247. My translation differs very slightly from that of Allsen, mainly stylistically. See also M.C. Brose, ‘People in the Middle: Uyghurs in the Northwest Frontier Zone,’ in ed. D.J. Wyatt, Battlefronts Real and Imagined: War, Border and Identity in the Chinese Middle Period (New York: Palgrave Macmillan, 2008), p.260; and M.C. Brose, ‘Uyghur Technologists of Writing and Literacy in Mongol China,’ T’oung pao 91 (2005): 396–435, at pp.405–06.

121 Morgan, ‘Persian as a Lingua Franca,’ pp.163–64.

122 L. Olschki, Marco Polo’s Asia: An Introduction to his ‘Description of the World’ Called ‘Il Milione’ (Berkeley: University of California Press, 1960), p.81.

123 On the Turkicisation of Central Asia, and the absorption by the Turks of other ethnic groups, who became Turkic speakers, see Golden, History of the Turkic Peoples, especially pp.152–53, on the Turks and trade; and pp.164–65, on the absorption by the Uighurs of eastern Iranians and Tokharians.

124 Y. Bregel, ‘Turko-Mongol Influences in Central Asia,’ in ed. R.L. Canfield, Turko-Persia in Historical Perspective (Cambridge: University Press, 1991), p.54; see p.53 for Bregel’s definition of ‘Central Asia’, which seems not to include modern Xinjiang.

125 Golden, History of the Turkic Peoples, pp.228–30, and Ch.12, pp.379ff.

126 Liu Xu 劉昫, et al., Jiu Tang shu 舊唐書 (Beijing: Zhonghua shuju, 1975),Vol.16, j.150, shang 上, p.5367; de la Vaissière, Sogdian Traders, pp.215–16.

127 I. Yakubovich, ‘Marriage Sogdian Style,’ in eds H. Eichner et al., Iranistik in Europa – gestern, heute, morgen (Wien: Österreichischen Akademie der Wissenschaften, 2006), pp.311–16; P. Lurje, ‘Once More on Sogdian pyšn’m’k ‘Surname’ and a Bridegroom Named ‘Hail’,’ in eds M. Ritter, R. Kauz and B. Hoffmann, Iran und iranisch gepr ä gte culturen: Studien zum 65. Geburtstag von Bert G. Fragner (Wiesbaden: Dr Ludwig Reichert Verlag, 2008), pp.232–33.

128 Y. Yoshida, ‘Turco-Sogdian Features,’ in eds W. Sundermann, A. Hintze and F. de Blois, Exegisti Monumenta: Festschrift in Honour of Nicholas Sims-Williams (Wiesbaden: Harrassowitz, 2009), pp.572–73.

129 M. Biran, The Empire of the Qara Khitai in Eurasian History: Between China and the Islamic World (Cambridge: University Press, 2005), pp.77–78.

130 ‘The collapse of the native Iranian dynasties of the north-east [in the tenth to eleventh centuries] was followed within a few decades by a major migration of Turkish peoples …’: C.E. Bosworth, ‘The Political and Dynastic History of the Iranian World (A.D. 1000–1217),’ in ed. J.A. Boyle, The Cambridge History of Iran, Vol.5, The Saljuq and Mongol Periods (Cambridge: University Press, 1968), p.9.

131 John of Plano Carpini, ‘History of the Mongols,’ in The Mongol Mission, (London: Sheed and Ward, 1955), p.59; original Latin text in Recueil de voyages et de m émoires , publiées par la Société de Géographie, Vol.4 (Paris: Arthus-Bertrand, 1839), p.750. It may be noted that this is clear evidence that there were many Muslim Turks at this time.

132 E. Çoban, ‘Eastern Muslim Groups among Hungarians in the Middle Ages,’ Bilig 63 (2012): 55–78, at pp.62–63; F. Hirth and W.W. Rockhill, Chau Ru-kua: His Work on the Chinese and Arab Trade in the Twelfth and Thirteenth C enturies, Entitled Chu-Fan-Chï (St. Petersburg: Imperial Academy of Sciences, 1911), p.16n.

133 John of Plano Carpini, ‘History of the Mongols,’ p.59; Recueil de voyages, Vol.4, p.750.

134 This is Dawson’s suggestion, in John of Plano Carpini, ‘History of the Mongols,’ p.59n.

135 Juvainī, History of the World Conqueror, Vol.1, p.83n.

136 Useful maps can be found in Y. Bregel, An Historical Atlas of Central Asia (Leiden: Brill, 2003), pp.35, 37.

137 On the presence of Turks in Transoxania, and even further to the south and west, as early as the seventh century, see R.N. Frye and A.M. Sayili, ‘Turks in the Middle East Before the Saljuqs,’ Journal of the American Oriental Society 63 (1943): 194–207, at pp.195–204; see also Y. Bregel, ‘Turko-Mongol Influences,’ p.60. Bregel states that Khwarazm was more or less completely Turkicised by the fourteenth century, and that the process must have started before the Mongol invasions.

138 William of Rubruck, ‘The Journey of William of Rubruck,’ in ed. Dawson, The Mongol Mission, p.137; original Latin text in Recueil de voyages et de m émoires, Vol.4, pp.281–82. The identification of Cailac with Qayalïq is strengthened by the fact that William says that the country of the Uighurs ‘adjoins the Organum territory’.

139 Morgan, ‘Persian as a Lingua Franca,’ p.165; William of Rubruck, ‘The Journey,’ p.136; original Latin text in Recueil de voyages et de m émoires, Vol.4, p.281.

140 Ibid., p.166; Pelliot, Notes, Vol.1, p.253. Pelliot does not express a clear preference for Quyas. As it was in the Ili region, east of Qayalïq, it seems unlikely that it was William’s ‘Equius’; more likely is Pelliot’s other choice, Iki-ögüz, south-west of Qayalïq (see B.D. Kotchnev, ‘Les frontières du royaume Karakhanide,’ Cahiers d’Asie Centrale 9 (2001): 41–48, at p.42). Morgan says that it was ‘well inside what is now Chinese Central Asia’, but, in fact, even the Ili region is very close to the border between Xinjiang, in China, and Kazakhstan. Most probably, however, ‘Equius’ was well outside what is today China.

141 A precise definition of ‘Persia’ still seems elusive today. According to Frye and Sayili, ‘Turks in the Middle East,’ p.201, the Shāhnāma took the Oxus (Amu Darya) to be the boundary between Persians and Turks in pre-Islamic times; see also K.H. Menges, The Turkic Languages and Peoples: an Introduction to Turkic Studies, 2nd ed. (Wiesbaden: Harrassowitz, 1994), p.29.

142 On Sogdian settlements in what is now Xinjiang, in the Gansu corridor, and even in northern China, see de la Vaissière, Sogdian Traders, pp.114–17, 122–44; and E.G. Pulleyblank, ‘A Sogdian Colony in Inner Mongolia,’ T’oung pao 41 (4/5) (1952): 317–56, at pp.326–28. Also, according to R.N. Frye, ‘Sughd and the Sogdians: A Comparison of Archaeological Discoveries with Arabic Sources,’ Journal of the American Oriental Society 63 (1943): 14–16, at p.16, ‘Mahmūd al-Kashgari in the eleventh century says that Sogdians had settled at Balāsāghūn, Tarāz, and Isfījāb, but Turkish was spoken in these towns’.

143 de la Vaissière says that there was Sogdian ‘colonization’ of Semirechye, towards the Issyk Kul; see Sogdian Traders, pp.114–16.

144 Ibid., pp.329–30, says that the language that had disappeared from the area must have been Sogdian, and that ‘Organum is certainly the land of Argu’.

145 Moule and Pelliot, Marco Polo, Vol.1, p.447.

146 Ibid., p.452.

147 Haw, Marco Polo’s China, p.86.

148 Clauson, Pre-Thirteenth-Century Turkish, p.205.

149 Morgan, ‘Persian as a Lingua Franca,’ p.163.

150 IDP Database search results. Available online at: <;random=7366>, viewed 7 Aug. 2013; see also N. Sims-Williams, ‘Early New Persian in Syriac script: Two Texts from Turfan,’ Bulletin of the School of Oriental and African Studies 74 (2011): 353–74.

151 R.L. Canfield, ‘Introduction: the Turko-Persian Tradition,’ in ed. Canfield, Turko-Persia in Historical Perspective, p.9; on the boundaries of the Qarakhanid state at various times, see Kotchnev, ‘Les frontières du royaume des Karakhanides,’ pp.41–48.

152 V.N. Nastich, ‘Persian Legends on Islamic Coins: From Traditional Arabic to the Challenge of Leadership,’ in eds B. Callegher and A. D’Ottone, The 2nd Simone Assemani Symposium on Islamic Coins (Trieste: Edizioni Università di Trieste, 2005), pp.170–71.

153 Morgan, ‘Persian as a Lingua Franca,’ pp.164–65. Morgan wonders if the seal indicated ‘that the Persian was the “official” version’. I think that all versions of the letter issued by the Qa’an’s secretariat would have carried such a seal impression, without which they would have been regarded as of no validity.

154 John of Plano Carpini, ‘History of the Mongols,’ p.67.

155 Findley, Turks in World History, pp.56–57, 72; Golden, History of the Turkic Peoples, p.224. On contacts between the Cumans and the Latins of Constantinople in 1237 and 1241, see I. Vásáry, Cumans and Tatars: Oriental Military in the Pre-Ottoman Balkans, 1185–1365 (Cambridge: Cambridge University Press, 2005), pp.63–66.

156 V. Spinei, The Romanians and the Turkic Nomads North of the Danube Delta from the Tenth to the Mid-Thirteenth Century (Leiden: Brill, 2009), p.154; V. Spinei, ‘The Cuman Bishopric – Genesis and Evolution,’ in ed. F. Curta, The Other Europe in the Middle Ages: Avars, Bulgars, Khazars and Cumans (Leiden: Brill, 2008), pp.423–26.

157 John of Plano Carpini, ‘History of the Mongols,’ p.56.

158 Fragner, ‘Shah Ismail’s Fermans,’ p.36.

159 William of Rubruck, ‘The Journey,’ p.124.

160 Ibid., p.209.

161 Ibid., p.208.

162 Matthew of Edessa (trans. A.E. Dostourian), Armenia and the Crusades, Tenth to Twelfth Centuries: The Chronicle of Matthew of Edessa, (Lanham: University Press of America, 1993), p.228. Cf. Matthieu d’Édesse, Chronique, (962–1136) (Paris: Durand, 1858), p.306. Dostourian, p.285n, comments that: ‘Whenever Matthew refers to the ‘Persians’, he means those Muslims inhabiting and ruling the Iranian plateau … . Therefore, the term ‘Persian’ can refer to both Turks and Iranians … .’

163 Findley, Turks in World History, p.68; C.E. Bosworth, ‘The Political and Dynastic History of the Iranian World (A.D. 1000–1217),’ in ed. J.A. Boyle, The Cambridge History of Iran, Vol.5, The Saljuq and Mongol Periods (Cambridge: Cambridge University Press, 1968), pp.1–202, gives a fairly concise account of the complex history of Persia at this period.

164 Bo Utas, ‘A Multiethnic Origin of New Persian?’ in eds L. Johanson and C. Bulut, Turkic-Iranian Contact Areas: Historical and Linguistic Aspects (Wiesbaden: Harrasso-witz, 2006), p.249.

165 A. Kasravī (trans. E Siegal), ‘The Turkish Language in Iran,’ Journal of Azerbaijani Studies 1 (1998): 45–73, at p.49; see also R. Levy, The Persian Language (New York: Philosophical Library, 1951), pp.8–9; and Menges, Turkic Languages and Peoples, pp.12–13.

166 J. Pfeiffer, ‘Confessional Ambiguity vs. Confessional Polarization: Politics and the Negotiation of Religious Boundaries in the Ilkhanate,’ in ed. J. Pfeiffer, Politics, Patronage and the Transmission of Knowledge in 13th–15th Century Tabriz (Leiden: Brill, 2014), p.131.

167 P. Pelliot, ‘Les Mongols et la Papauté (1),’ Revue de l’Orient Chrétien, 3.23 (1922–23): 1–30, at pp.18, 24–25; for an English translation, see I. de Rachewiltz, Papal Envoys to the Great Khans (London: Faber, 1971), pp.213–14. The translator, J.A. Boyle, notes that ‘the preamble is in Turkish’.

168 Moule and Pelliot, Marco Polo, Vol.1, p.78; Yule, Marco Polo, 3rd ed., Vol.1, p.13, gives ‘Tartar tongue’, but it seems that most manuscripts say ‘Turkish’.

169 Haw, Marco Polo’s China, p.49.

170 On Chinqai, see de Rachewiltz, et al., eds., In the Service of the Khan, pp.95–111.

171 This is roughly correct for Uighur script, but too few for Arabic.

172 Peng Daya 彭大雅 and Xu Ting 徐霆, Heida s hilue jianzheng 黑韃事略箋證, ed., Wang Guowei 王國維, in Wang Guowei yishu 王國維遺書, Vol.8 (Shanghai: Shanghai shudian, 1983), pp.211–12. This passage (and more) is also translated by F.W. Cleaves, ‘A Chancellery Practice of the Mongols in the Thirteenth and Fourteenth Centuries,’ Harvard Journal of Asiatic Studies 14 (1951): 493–526, at pp.497–503, with annotations.

173 Morgan, ‘Persian as a Lingua Franca,’ p.166; I. de Rachewiltz, ‘Marco Polo Went to China,’ Zentralasiatische studien 27 (1997): 34–92, at p.55 and note; Huang Shijian, ‘The Persian Language in China During the Yuan Dynasty,’ Papers on Far Eastern History 34 (1986): 83–95; at p.94, Huang refers to conferring with Professor de Rachewiltz. The same claim is made by C. Salmon in ‘Les Persans à l’extrémité orientale de la route maritime’ (2 A.E.- 17 siècle,’ Archipel 68 (2004): 23–58, at p.24, again based at least partly on Huang’s work. Allsen also seems to have accepted Huang’s opinion; see T.T. Allsen, ‘The Ras û lid Hexaglot in its Eurasian Cultural Context,’ in ed. P.B. Golden, The King’s Dictionary: The Ras û lid Hexaglot (Leiden: Brill, 2000), p.37.

174 de Rachewiltz, ‘Turks in China under the Mongols,’ p.308n; on his change of opinion, see de Rachewiltz, ‘Marco Polo Went to China,’ p.55n.

175 Morgan, ‘Persian as a Lingua Franca,’ p.167.

176 In Yule’s translation: Marco Polo, 3rd ed., Vol.1, pp.16, 34, 35.

177 YS, Vol.1, j.3, p.51.

178 Pelliot, Notes, Vol.1, p.23; Leslie, Islam, pp.195–96; Liu Yingsheng, ‘A Lingua Franca along the Silk Road: Persian Language in China Between the 14th and the 16th Centuries,’ in ed. R. Kauz, Aspects of the Maritime Silk Road: From the Persian Gulf to the East China Sea (Wiesbaden: Harrasso-witz, 2010), pp.87–88.

179 M. Biran, ‘Qarakhanid Studies: A View from the Qara Khitai Edge,’ Cahiers d’Asie Centrale 9 (2001): 77–89, at pp.77–78; Biran notes that the Qarakhanids were ‘the first Muslim Turkic dynasty’, and that Chinese sources are important for Qarakhanid studies.

180 A. Waley, The Travels of an Alchemist: The Journey of the Taoist Ch’ang-Ch’un from China to the Hindukush at the Summons of Chingiz Khan (London: Headley Brothers, 1931), p.36; see also F.W. Cleaves, ‘A Chancellery Practice of the Mongols,’ p.501n.

181 A. Vissière, ‘Les Désignations Ethniques Houei-houei et Lolo,’ Journal Asiatique, 11.3 (1914): 175–82, at p.176; Liu Yingsheng, ‘A Lingua Franca,’ p.87; Leslie, Islam, p.196; Yang Zhijiu 楊志玖, Yuandai Huizu shigao 元代回族史稿 (Tianjin: Nankai daxue chubanshe, 2003), p.7.

182 For example, YS, Vol.10, j.121, p.2976.

183 I. de Rachewiltz, The Secret History of the Mongols: A Mongolian Epic Chronicle of the Thirteenth Century (Leiden: Brill, 2006), Vol.1, p.562.

184 O. Lattimore, ‘Introduction,’ in O. Lattimore and L.M.J. Schram, The Monguors of the Kansu-Tibetan Frontier: Their Origin, History, and Social Organization, (Philadelphia: American Philosophical Society, 1954), pp.6–7; see also Maria E. Subtelny, ‘The Symbiosis of Turk and Tajik,’ in ed. B.F. Manz, Central Asia in Historical Perspective (Boulder: Westview Press, 1998), p.49.

185 I. de Rachewiltz, The Secret History of the Mongols, Vol.1, p.194; Vol.2, pp.962–63. ‘Qurumši’ refers to Khwarazm. As already seen above, according to John of Plano Carpini, the Khwarazmians spoke ‘the Coman language’. See also n.194 below, for reference to biographies of Mahmūd and his descendants.

186 These translations are somewhat un certain. On types of fibres used for making cloth, see J. Needham and D. Kuhn, Science and Civilisation in China, Vol.5, Chemistry and Chemical Technology, pt.9, Textile Technology: Spinning and Reeling (Cambridge: Cambridge University Press, 1988), pp.15–59.

187 Zhou Mi 周密, Guixin za shi 癸辛雑識, xuji shang 續集上 (Beijing: Zhonghua shuju, 1988), pp.142–43. Leslie gives a translation of this passage in Islam, pp.67–68; it must be noted that there are a number of errors in his translation, some of them serious. He comments that: ‘In spite of one or two peculiar features, this is still an excellent description and mostly believable as an eye-witness account’. He clearly failed to realise that the ‘peculiar features’ were normal Turkic customs.

188 Qian Shuoyou 潛説友, Xianchun Lin’an zhi 咸淳臨安志 (Taibei: Chengwen chubanshe, 1970), Vol.1, j.13, pp.162–63.

189 D. Sinor, ed., The Cambridge History of Early Inner Asia (Cambridge: Cambridge University Press, 1990), p.197.

190 Sinor, Early Inner Asia, p.304.

191 C. Mackerras, The Uighur Empire According to the Tang Dynastic Histories (Canberra: ANU Press, 1972), p.68; the original Chinese text is in Jiu Tang shu, Vol.16, j.195, p.5202.

192 Hu Sanxing 胡三省, Zi zhi tong jian yin zhu 資治通鑑音註, in Sima Guang 司馬光, Zi zhi tong jian 資治通鑑 (Beijing: Zhonghua shuju, 1956), Vol.15, j.221, p.7076; this passage is paraphrased in Mackerras, The Uighur Empire, p.136n.

193 See A.C. Moule’s long note on ‘Quinsai’ in his Quinsai, With Other Notes on Marco Polo (Cambridge: Cambridge University Press, 1957), pp.1–53.

194 de Rachewiltz, et al., eds, In the Service, p.122.

195 See above, note 39; it may be noted that Sayyid Ajall’s grandfather (probably) surrendered to the Mongols ‘at the head of a thousand horsemen’ (de Rachewiltz, et al., In the Service, p.467), which perhaps suggests that he may have been of Turkic extraction. It was usually Turks who were the cavalry of Transoxania at this time.

196 A biography of Ahmad, by H. Franke, may be found in de Rachewiltz, et al., In the Service, pp.539–57: it is interesting to note that his biography in the YS, Vol.15, j.205, p.4558, originally described him as an Uighur, using the old term (more or less obsolete during the Yuan period) Huihe 回纥. This is changed by most editors to Huihui 回回 (see note 1, YS, Vol.15, p.4588), but it is quite likely that someone from Fanākat was both an Uighur and a Muslim.

197 Leslie, Islam, p.79; YS, Vol.10, j.122, p.3016; P. Pelliot, ’Une ville musulmane dans la Chine du nord sous les Mongols,’ Journal Asiatique 211 (1927): 261–79, at pp.261, 278–79.

198 Chen Dasheng 陳逹生, Quanzhou Yisilanjiao shi ke 泉州伊斯兰教石刻 (Fuzhou: Ningxia renmin, Fujian renmin, 1984), pp.15–40 (and pp.29–90 of the English translation).

199 T.T. Allsen, Culture and Conquest in Mongol Eurasia (Cambridge: Cambridge University Press, 2001), pp.41–42; M. Rossabi, ‘The Mongols and their Legacy,’ in eds. S. Carboni and L. Komaroff, The Legacy of Genghis Khan: Courtly Art and Culture in Western Asia, 1256–1353 (New York: Metropolitan Museum of Art, 2002), p.24.

200 See de la Vaissière, Sogdian Traders, pp.306–09, 311, 325–26, 328–30; Golden, History of the Turkic Peoples, pp.172–73, 197–99.

201 Tombstones are, in fact, not a very good reflection of the living population. Some who died in a certain place may have been taken back to their home area for burial. Others may have returned before death, when they became old or infirm.

202 For example, Leslie, Islam, p.86.

203 Chin-Fu Hung, ‘China and the Nomads: Misconceptions in Western Historiography on Inner Asia,’ Harvard Journal of Asiatic Studies 41 (1981): 597–628, at p.617.

204 Wu Wenliang 吴文良, Quanzhou zongjiao shike 泉州宗教石刻 (Beijing: Kexue chubanshe, 1957). About half of Wu’s book is devoted to Islamic inscriptions and half to those of other religions, but by no means all date from the Yuan period, of course.

205 Pelliot, Notes, Vol.2, p.727. It may be noted that this proscription extended to all foreign religions, including Islam.

206 Rossabi, Khubilai Khan, p.13.

207 S.N.C. Lieou, ‘Nestorians and Manichaeans on the South China Coast,’ Vigiliae Christianae 34 (1980): 71–88, at p.73.

208 Niu Ruji 牛汝极, Shi zi lianhua: Zhongguo Yuandai Xuliyawen Jingjiao beiming wenxian y anjiu 十字蓮花: 中國元代叙利亚文景教碑銘文献研究 (Shanghai: Guji chubanshe, 2008), pp.1–41, at p.8.

209 Wang Yun, Qiujian xiansheng daquan wenji, j.88, p.5b. The text actually says: ‘Zuo maimai Weiwuer, Musuerman, Huihui … 做買賣畏吾兒木速兒蠻回回 …’; perhaps here ‘Musuerman’, Musulman, means Persian Muslims, while ‘Huihui’ refers to Turkic Muslims, the combination of the two meaning Muslims in general. A translation of this passage may be found in E. Endicott-West, ‘Merchant Associations in Yüan China: The Orto ɣ,’ Asia Major 2.2 (1989): 127–54, pp.142–43; but her translation ‘the Uighurs who are engaged in trade [and] the Muslims’ should, I think, be corrected to ‘the Uighurs [and] the Muslims who are engaged in trade’.

210 For a discussion of merchants during the Yuan period, see Chen Gaohua 陳高華 and Shi Weimin 史衛民, Zhongguo j ingji tongshi: Yuandai jingji juan 中國經濟通史 : 元代經濟卷 (Beijing: Jingji ribao chubanshe, 2000), pp.457–65; and also Yokkaichi Yasuhiro, ‘Chinese and Muslim Diasporas and the Indian Ocean Trade Network under Mongol Hegemony,’ in ed. A. Schottenhammer, The East Asian ‘ Mediterranean’: Maritime Crossroads of Culture, Commerce and Human Migration (Wiesbaden: Harrassowitz, 2008), pp.73–102, especially pp.88–90.

211 J.W. Christie, ‘The Medieval Tamil-language Inscriptions in Southeast Asia and China,’ Journal of Southeast Asian Studies, 29 (1998): 239–68, at p.266.

212 For example, C.P. Atwood, Encyclopedia of Mongolia, p.37: ‘Bayan tried to revive the old ethnic hierarchy’; M.C. Brose, ‘Realism and Idealism in the Yuanshi Chapters on Foreign Relations,’ Asia Major, 19.1 (2005): 327–47, at p.345: ‘they [the Mongols] categorized other peoples according to ethnographic terms’; and also Hung, ‘China and the Nomads,’ pp.624–25: ‘As is well known to both Chinese traditional historians and modern scholars, Yüan society as a whole discriminated among populations of different ethnic and geographic origins’. ‘Geographic’ is perhaps not completely unjustified here, but it implies an emphasis that is not, in my view, correct. The difference between Han ren and Nan ren was not based on geography, despite my translations, but on when the different groups had come under Mongol rule.

213 The Tangut region, that is, the former Xi Xia 西夏 state.

214 YS, Vol.2, j.13, p.268.

215 On the Qara Khitai empire, see Biran, The Empire of the Qara Khitai.

216 The Hexi region had formed a large part of the former Tangut, or Xi Xia, state. It may be noted that there were a significant number of Uighurs in the Hexi region.

217 de Rachewiltz, ‘Turks in China under the Mongols,’ pp.282, 297n.

218 Tao Zongyi 陶宗儀 lists 31 different kinds of Semu ren in Nancun chuo geng lu 南村輟耕錄 (Beijing: Zhonghua shuju, 1959), j.1, p.13; for a discussion of his list, see Yanai Wataru 箭内互 (trans. Chen Jie 陳捷 and Chen Qingquan 陳清泉), Yuandai Meng Han s emu daiyu kao 元代蒙汉色目待遇考 (Shanghai: Shangwu yinshuguan, 1932), pp.18–29.

219 The following counts were obtained using a searchable electronic version of the text of the Yuan shi.

220 There were more groups of non-Muslim Semu ren than I have included in these counts.

221 de Rachewiltz, ‘Turks in China under the Mongols’.

222 Ibid., pp.287, 293.

223 Ibid., p.285, Table 10.1.

224 Leslie, Islam, pp.102–03.

225 For example, among Leslie’s 43 is Mahmud Yalavach; it must also be remarked that he includes Ai Xie 爱薛, who, as he himself notes, was certainly a Christian, and a few others who were probably not Muslims.

226 de Rachewiltz, et al., In the Service, pp.v–vi. Turks are clearly represented, however.

227 Liu Yingsheng, ‘A Lingua Franca,’ p.89.

228 Mongolian appears twice, written with the 'Phags-pa script and also with the Uighur script.

229 Liu Yingsheng, ‘A Lingua Franca,’ p.89n.

230 YS, Vol.15, j.202, p.4518; Coblin, Handbook of 'Phags-pa Chinese, pp.1–4; Allsen, ‘The Rasûlid Hexaglot,’ p.26.

231 Chen Yongzhi 陳永志, ‘Meng Yuan shiqide paifu 蒙元時期的牌符,’ Nei Menggu Daxue Xuebao (Renwen shehui kexue ban), XXXV (2003): 30–37.

232 Hugejiletu Sarula 呼格吉勒圖 薩如拉, Basibazi Mengguyu wenxian huibian 八思巴字蒙古語文献匯編 (Huhehaote: Nei Menggu jiaoyu chubanshe, 2004), pp.455–74.

233 Li Aiping 李爱平 and Chen Yongzhi 陳永志, ‘Wulan Chabushi “Guo Bao” daojue xu: zhuanjia jiedu Mengyuan paifu’ (8/5/2008). Available online at: < Text/2008050817979.html>, viewed 8 Feb. 13. The list is basically the same as in Chen Yongzhi’s article cited above, note 219, with one addition.

234 D.P.Leidy, Wai-fong A. Siu and J.C.Y. Watt, ‘Chinese Decorative Arts,’ The Metropolitan Museum of Art Bulletin 55.1 (1997): 1–71, at p.9. This paizi carries an inscription in Mongolian only, written with the 'Phags-pa script.

235 Jakhadai Chimeddorji 齊木德道尔吉, ‘Xinan Daxue Lishi Bowuguan cang Yuandai Mengguyu Basibawen paifu jiedu’ 西南大學歷史博物館藏元代蒙古語八思巴文牌符解讀, Mengguxue jikan 蒙古学 辑刊 17 (2007): 1–13. These two paizi bear inscriptions only in Mongolian written with the 'Phags-pa script.

236 English-language articles published in 2001 and 2003 list a similar number of paizi; Dang Baohai, ‘The Paizi of the Mongol Empire,’ Zentralasiatische studien 31 (2001): 31–62; Dang Baohai, ‘The Paizi of the Mongol Empire (continued),’ Zentralasiatische studien 32 (2003): 7–10.

237 Huang Shijian, ‘Persian language in China,’ p.93.

238 By way of comparison, it has been suggested that the use of a Chinese character on a coin struck in Khotan under Qaidu was a declaration of ‘the legality of his claims to power’; V.A. Belyaev and S.V. Sidorovich, ‘A New Interpretation of the Character Bao on Coins of Mongol Uluses,’ Journal of the Oriental Numismatic Society 199 (2009): 11–17, at pp.16–17.

239 Biran, The Empire of the Qara Khitai, p.93; V.A. Belyaev and S.V. Sidorovich, ‘Ji’erjisi faxiande “Xuxing yuan bao” yu Xi Liao de nianhao kao’ 吉尔吉斯發現的 “續興元寶”與西遼年號考, Zhongguo qianbi/China numismatics 116 (2012.1): 70–74 [illustrations on back cover]; V. Belyaev, V. Nastich and S. Sidorovich, ‘The Coinage of Qara Khitay: a New Evidence (On the reign-title of the Western Liao Emperor Yelü Yilie),’ in eds B. Callegher and A. D’Ottone, 3rd Simone Assemani Symposium on Islamic Coins (Trieste: Edizioni Università di Trieste, 2012), pp.130, 138, 143.

240 S.V. Sidorovich, ‘Zhu you Hanzide Buhuala qianbi xin kao’ 鑄有漢字的不花剌錢币新考, Zhongguo qianbi/China Numismatics 100 (2008.1): 21–24.

241 V.A. Belyaev and S.V. Sidorovich, ‘Dazhi you Hanzide Jin Zhang Han Guo Zhuchi yin bi’ 打制有漢字的金帳汗國术赤銀币, Zhongguo qianbi/China Numismatics 123 (2013.4): 23–30.

242 Belyaev and Sidorovich, ‘The Character Bao on Coins of Mongol Uluses,’ pp.11–16.

243 Huang Shijian 黄時鑒, ‘Yuandai siti mingwen tongquande kaoshi 元代四體銘文銅權的考釋,’ in ed. Ye Yiliang 葉奕良, Yilangxue z ai Zhongguo lunwen ji 伊朗學在中國論文集, (2) (Beijing: Beijing daxue chubanshe, 1998), pp.42–43.

244 V. Rybatski, ‘Between East and West: Central Asian Writing Systems,’ in eds M. Ölmez & F. Yıldırım, Orta Asya’dan Anadolu’ya Alfabeler (Istanbul: EREN, 2011), pp.215–16, 216n.

245 M. Erdal, ‘The Turkish Yarkand Documents,’ Bulletin of the School of Oriental and African Studies 47 (1984): 260–301, at pp.291–92, 296. It is perhaps relevant to my thesis to note that these documents from Yarkand, dating to its time under Qarakhanid rule, formed part of a collection of papers in Arabic and Uighur Turkic, but not Persian; Erdal, ‘The Turkish Yarkand Documents,’ p.261.

246 Liu Youzheng 劉幼錚, ‘Jieshao Tianjin faxiande yipi tong, tie quan’ 介紹天津 發現的一批銅,鐵權, in Wenwu ziliao congkan 文物資料叢刊 8 (Beijing: Wenwu chubanshe, 1983), pp.113–15.

247 Huang Shijian, ‘Yuandai siti mingwen tongquan,’ pp.43–45. Huang claims to be able to read only four of the ‘Persian’ inscriptions in full: the others are apparently in poor condition and partly or wholly illegible.

248 Liu Yingsheng, ‘A Lingua Franca,’ pp.87–95.

249 Ibid., p.88.

250 There is, in fact, a certain amount of evidence, but it appears that foreign vocabulary in the language of the modern Hui is as much Arabic and Turkic as Persian; see M. Dillon, China’s Muslim Hui Community: Migration, Settlement and Sects (Richmond: Curzon, 1999), pp.154–55.

251 Muriel Atkin, ‘Tajiks and the Persian World,’ in ed. Beatrice F. Manz, Central Asia in Historical Perspective (Boulder: Westview Press, 1998), p.131, says that ‘New Persian …, evolved as a literary language in the region from Central Asia to Sistan … in the ninth and tenth centuries …’ [emphasis added].

252 I. Silova and T. Abdushukorova, ‘Global Norms and Local Politics: Uses and Abuses of Education Gender Quotas in Tajikistan,’ Globalisation, Societies and Education 7 (2009): 357–76, at p.361.

253 For a description of this paizi, and reproductions of rubbings of its two faces, see Hugejiletu Sarula, Basibazi Mengguyu Wenxian huibian, pp.473–74, and plate 47.

254 Cai Meibiao 蔡美彪 et al., ‘Dui Keyou Zhongqi yexun pai alabo zimu wenzi dushide yijian,’ 對科右中旗夜巡牌阿拉伯字母文字讀釋的意見 Minzu yuwen 民族語文 1995(3): 51–55, at p.51.

255 Cai Meibiao et al., ‘Keyou Zhongqi yexun pai,’ p.54. Liu here seems to contradict what he says in ‘A Lingua Franca,’ pp.92–95, regarding Persian in China during the Ming dynasty.

256 Hao Sumin 郝蘇民 and Liu Wenxing 劉文性, ‘Guanyu Keyou Zhongqi yexunpai alabo zimu wenzi dushide zai taolun,’ 関于科右中旗夜巡牌阿拉伯字母文字讀釋的再討論 Minzu yuwen 1996(3): 71–72.

257 Cai Meibiao et al., ‘Keyou Zhongqi yexun pai,’ p.53.

258 G. Doerfer, Türkische und Mongolische elemente im Neupersischen unter besondere Berüchsichtigung älterer neupersischer Geschichtsquellen, vor Allem der Mongolen- und Timuridenzeit, Band 1, Mongolische elemente im Neupersischen, (Wiesbaden: Steiner, 1963), pp.239–41.

259 Erdal, ‘The Turkish Yarkand Documents,’ p.265, notes that the vocabulary of Turkic documents dating from the late eleventh to early twelfth centuries is ‘full of Arabic and New Persian elements’.

260 Allsen, ‘The Rasûlid Hexaglot,’ p.38.

261 E. Endicott-West, Mongolian Rule in China: Local Administration in the Yuan Dynasty (Cambridge and London: Council on East Asian Studies, Harvard University, and Harvard-Yenching Institute, 1989), pp.83–84.

262 A. Göksel and C. Kerslake, Turkish: A Comprehensive Grammar (London: Routledge, 2005), pp.144–45; M. Erdal, A Grammar of Old Turkic (Leiden: Brill, 2004), pp.380–85. On p.383, Erdal notes an interesting case of a Turkic translation from Parthian influenced by Parthian word-order, which is clearly different from usual Turkic word-order.

263 It may be noted that Hugejiletu Sarula says that the inscription is Chaghatai; see Basibazi mengguyu wenxian huibian, p.473.

264 See, for example, R. Gluckman, ‘Remade in China: The Fine Art of Fakery’ (2002), < html>, viewed 19 Jan. 2013, and Li Qian, ‘Should fake antique dealers be punish-ed?,’ China Daily (27/2/2007), < 27/content_814924.htm>, viewed 19 Jan 2013. There is also an issue of Extrême orient, extrême occident devoted to ‘Faux et falsification en Chine, au Japon, et au Viêt Nam’ 32 (2010).

265 M. Balter, ‘Authenticity of China’s Fabulous Fossils Gets New Scrutiny,’ Science 340 (2013): 1153–54; X. Wang, ‘Mortgaging the Future of Chinese Paleontology,’ Proceedings of the National Academy of Sciences of the United States of America 110.9 (2013): 3201.

266 Usually, all that is said is that a certain item was found in a certain place in a certain year, and nothing more; see, for example, Hugejiletu Sarula, Basibazi Mengguyu wenxian huibian, pp.455–73. Interestingly, it seems that there is also often a lack of published information about the provenance of fossils from China; Balter, ‘Authenticity of China’s Fabulous Fossils,’ p.1154.

267 This is the first paizi described in Hugejiletu Sarula, Basibazi Mengguyu wenxian huibian, pp.455–57.

268 Secrecy would have been essential, at least during the period of the Cultural Revolution in the late 1960s and early 1970s.

269 ‘Chutu Yuandai shengzhi jin pai’ 出土元代聖旨金牌 (2010), <http://www.>, viewed 13 Jan. 2013.

270 A. Vardanyan, ‘Some Additions to the Coins with the Inscription ‘Ulugh Mangyl Ulūs (Ulūsh) Bek’,’ Journal of the Oriental Numismatic Society 190 (2007): 7–20; Badarch Nyamaa, The Coins of the Mongol Empire and Clan Tamgha of Khans (XIII–XIV), (Ulaan Baataar: the author, 2005), pp.44–45.

271 Vardanyan, ‘Coins With the Inscription “Ulugh Mangyl Ulūs (Ulūsh) Bek”,’ pp.13, 19–20.

272 Nastich, ‘Persian Legends on Islamic Coins,’ pp.173–74.

273 Badarch Nyamaa, Coins of the Mongol Empire, passim. (Mongolian in Uighur script, and Arabic); p.106 (inscription in 'Phags-pa from the Jochid Ulus); p.114 (Yuan coins with 'Phags-pa inscriptions); pp.124–27 (Chinese inscriptions);.p.139 (inscription in 'Phags-pa from the Ilkhanate); p.135 (coin of Ögödei Qa’an with Arabic and Persian inscriptions); p.137 (Persian inscriptions from the Ilkhanate).

274 Nastich, ‘Persian Legends on Islamic Coins,’ p.179.

275 Ibid., p.180.

276 Liu Yingsheng, ‘A Lingua Franca,’ p.92.

277 This is the translation of Farquhar, Government of China, pp.44, 158n. It was also adopted by Rossabi, ‘The Mongols and their Legacy,’ p.26.

278 YS, Vol.2, j.15, p.322; Vol.7, j.81, pp.2028–29. There are other references to the Muslim National Institute and to officials who taught the ‘Istifi’ script, as will be seen shortly.

279 There are not many of them, but I have found mention of the Muslim National College in a short piece by Huang Jin 黄溍, ‘Hanlin guoshi yuan timing ji’ 翰林國史院題名記, in ed. Li Xiusheng 李修生, Quan Yuan wen 全元文, Vol.29 (Nanjing: Jiangsu guji chubanshe, 2004), j.952, pp.297–98. There is also a brief mention in the Tongzhi tiaoge; Fang Linggui 方齡貴, ed., Tongzhi tiaoge j iaozhu 通制條格校注 (Beijing: Zhonghua shuju, 2001), j.5, pp.247–48.

280 The expression used is usually Yisitifei wenzi 文字, which unambiguously means ‘script’.

281 See, for example, Mozafar Bakhtyar, ‘Yisitifei kao’ 亦思替非  考, in ed. Ye Yiliang 葉奕良, Yilangxue zai Zhongguo lunwen ji 伊朗學在中國論文集 (1) (Beijing: Beijing daxue chubanshe, 1993), pp.44–50; a more reasonable view was expressed by Han Rulin 韓儒林, ‘Suo wei “Yisitifei” wenzi shi shenma wenzi?’ 所謂‘亦思替非’ 文字是甚麽文字?, in his collected writings, Qiong lu ji 穹廬集 (Shijiazhuang: Hebei jiaoyu chubanshe, 2000), pp.292–94. However, he too fails to distinguish clearly between language and script.

282 O. Leaman, ed., The Qur’an: An Encyclopedia (New York: Routledge, 2006), p.324.

283 Both Leslie, Islam, p.95, and Farquhar, Government of China, p.158n, took it to mean the Arabic script.

284 Liu Yingsheng, ‘A Lingua Franca,’ p.92.

285 YS, Vol.7, j.81, p.2028–29.

286 Farquhar, Government of China, p.44.

287 This certainly means a ‘textual translator’. See Allsen, ‘The Ras û lid Hexaglot,’ p.30.

288 YS, Vol.7, j.81, p.2029.

289 Hyunhee Park, Mapping the Chinese and Islamic Worlds: Cross-Cultural Exchange in Pre-modern Asia (Cambridge: Cambridge University Press, 2012), p.99.

290 Morgan, ‘Persian as a Lingua Franca,’ p.166.

291 The translation ‘National Institute’ follows Farquhar, Government of China, p.130.

292 YS, Vol.2, j.25, p.565; YS, Vol.7, j.81, p.2028; YS, Vol.3, j.27, p.601. D.D. Leslie, ‘Living With the Chinese: The Muslim Experience in China, T’ang to Ming,’ in eds C. Le Blanc and S. Mader, Chinese Ideas about Nature and Society: Studies in Honour of Derk Bodde (Hong Kong: Hong Kong University Press, 1987), p.182, gives the date as 1321, and says that both the Institute and the College were abolished, but these are errors.

293 Liu Yingsheng, ‘A Lingua Franca,’ p.92.

294 Liu apparently took the character fu 复 to mean ‘again’, which is possible, but in this case it clearly means ‘in addition’; the National Institute was not re-established, but was established in addition to the National College. There is nowhere any mention of a Muslim National Institute having existed previously.

295 Dawson, Mongol Mission, p.228.

296 The Nestorian church was, of course, based in Persia; see F. Micheau, ‘Eastern Christianities (Eleventh to Fourteenth Century): Copts, Melkites, Nestorians and Jacobites,’ in ed. M. Angold, The Cambridge History of Christianity, Vol.5, Eastern Christianity (Cambridge: Cambridge University Press, 2006), pp.377–78.

297 For the second variant, see A. Schuessler, Dictionary of Old Chinese, p.309.

298 Wang Gong 王鞏, Jiashen zaji 甲申雜記, in Zhongguo yeshi jicheng 中國野史集成 (Chengdu: Sichuan daxue tushuguan, 1993), Vol.8, p.458.

299 Rossabi, ‘The Mongols and Their Legacy,’ p.26.

300 P.D. Buell and E.N. Anderson, A Soup for the Qan: Chinese Dietary Medicine of the Mongol Era as Seen in Hu Sihui’s Yinshan zhengyao, 2nd ed. (Leiden: Brill, 2010), p.65.

301 Buell and Anderson, Soup for the Qan, pp.106–13.

302 J. Needham, Science and Civilisation in China, Vol.3, Mathematics and the Sciences of the Heavens and the Earth (Cambridge: Cambridge University Press, 1959), p.49; Ibid.,Vol.6, Biology and Biological Technology, pt6, Medicine (Cambridge: Cambridge University Press, 2000), p.53; Liu Yingsheng, ‘A Lingua Franca,’ pp.89–92 (but it must be noted that several of the terms Liu lists as though they were Persian are, in reality, Arabic). Allsen, Culture and Conquest, p.151, states that Huihui medicine ‘was almost always in the hands of Nestorians’; Nestorians often had some knowledge of Persian.

303 H. Franke and D. Twitchett, eds, The Cambridge History of China, Vol.6, Alien Regimes and Border States, 907–1368 (Cambridge: Cambridge University Press, 1994), p.537.

304 Rossabi, ‘Muslims in the Early Yuan,’ pp.261–62; F.W. Cleaves, ‘The Historicity of the Baljuna Covenant,’ Harvard Journal of Asiatic Studies, XVIII (1955): 357–421, at pp.396–97, 403.

305 Leslie, Islam, p.88; Leslie, ‘Living With the Chinese’, pp.181–82.

306 Da Yuan Shengzheng Guochao Dianzhang [facsimile of Yuan ed.] (Beijing: Zhongguo guangbo dianshi chubanshe, 1998), Vol.3, j.57, pp.2057–58; the translation follows Leslie, ‘Living With the Chinese,’ p.181. See also P. Jackson, ‘The Mongols and the Faith of the Conquered,’ in eds R. Amitai and M. Biran, Mongols, Turks, and Others: Eurasian Nomads and the Sedentary World (Leiden: Brill, 2005), pp.260–61.

307 This may actually mean tattooing; see C. Reed, ‘Tattoo in Early China,’ Journal of the American Oriental Society 120 (2000): 360–76, at pp.365–66.

308 YS, Vol.2, j.16, p.339.

309 Ibid., pp.364, 371.

310 YS, Vol.1, j.11, p.231; YS, Vol.2, j.16, p.348.

311 YS, Vol.3, j.27, p.607.

312 YS, Vol.3, j.27, p.611.

313 Leslie, Islam, pp.89–90.

314 Wu Hanchen [attrib.], ‘Yu hu chun,’ in ed. Zang Jinshu, Yuan qu xuan (Beijing: Zhonghua shuju, 1958), Vol.2, p.481. ‘Jurchen’ here translates the original Lulu 虜虜. Lu was commonly used by subjects of the Southern Song to refer, pejoratively, to Jurchens of the Jin empire, so I have taken this to be the usage of Lulu here. On the comparable, but earlier, Northern Song use of lu to refer to Khitans, see Tao Jing-shen, ‘Barbarians or Northerners: Northern Sung Images of the Khitans,’ in Rossabi, China Among Equals, p.72 (but note that some of Tao’s conclusions are invalidated by his failure to realise that texts in the Si ku quan shu 四庫全書 have been edited to remove pejorative references to Khitans and, especially, Jurchens, because of Manchu sensibilities). See also A. Vissière, ‘Les Désignations Ethniques Houei-houei et Lolo,’ pp.175–82 ; but Vissière is almost certainly wrong to identify Lulu with Lolo.

315 Tao Zongyi, Nancun chuogeng lu, j.28, p.348. This passage is translated by Leslie, Islam, p.93.

316 Golden, History of the Turkic Peoples, p.292.

317 Ibid., pp.292–93.

318 R. Finch, ‘Christianity among the Cumans,’ Surugadai University Studies 35 (2008): 75–96, at pp.78–79.

319 J.A. Boyle, ‘Rashīd al-Dīn and the Franks,’ Central Asiatic Journal, 14 (1970): 62–67, at p.67; D.O. Morgan, ‘The Mongol Empire: A Review Article,’ Bulletin of the School of Oriental and African Studies 44 (1981): 120–25, at p.122.

320 I. de Rachewiltz, ‘The Genesis of the Name “Yeke Mongγol Ulus”,’ East Asian History 31 (2006): 53–56, at p.54.

321 See L. Hambis, ‘L’histoire des Mongols avant Gengis-Khan d’après les sources chinoises et mongoles, et la documentation conservée par Rašīdu-d-Dīn,’ Central Asiatic Journal 14 (1970): 125–33, especially p.126.

322 Decades ago, V.V. Barthold realised this: see his Turkestan down to the Mongol Invasion, 2nd ed., (London: Luzac, 1928), p.44; but Barthold was apparently aware only of the Yuan shi, and there are many other Chinese sources. Also, J. Fletcher, ‘The Mongols: Ecological and Social Perspectives,’ Harvard Journal of Asiatic Studies XLVI (1986): 11–50, at p.27, referred to ‘the Persian histories and the more voluminous but less exploited material in Chinese …’. Judith Kolbas’ reference to ‘the area [of the Mongol Empire] that has the most source material, Greater Iran …’ can only be ascribed to ignorance; J. Kolbas, Review of Early Mongol Rule in Thirteenth Century Iran: A Peasant [sic, correctly ‘Persian’] Renaissance by George Lane, Journal of the Royal Asiatic Society 15.2 (2005): 243–46, at p.243.

323 See S.G. Haw, ‘Bayan of the Bārin’s Persian Wife, and Other Perplexities,’ Journal of Asian History (forthcoming), for an argument that Rashīd al-Dīn’s information about China, at least, is often defective and unreliable.

324 K. Jahn, ‘Rashīd al-Dīn and Chinese Culture,’ Central Asiatic Journal 14 (1970): 134–47, at p.140.