Here is an attempt to translate into English de la Vaissière’s very important 2005 paper, which establishes the equivalence of Huns and Xiongnu once and for all:
“Huns et Xiongnu,” CAJ, 2005
Étienne de la Vaissière (EPHE, Paris)
I. Huns, Xwn and Xiongnu
1. Maenchen-Helfen and the birth of a scientific field
It has been a quarter of a millennium since de Guignes proposed to identify Xiongnu with the Huns in his General History of the Huns, Turks, Mongol Tatars and other Western Tatars (1756–8). A considerable historiography had been built on this assumption since then until Maenchen-Helfen, in a series of articles fifty years ago, took it down completely and firmly stated the lack of link between Xiongnu and Huns, temporarily putting an end to the debate. Maenchen-Helfen’s articles are at the root of almost all of the current historiography. Served by his unrivalled knowledge of Soviet archaeology, Maenchen-Helfen succeeded in making the history of European Huns an autonomous domain. All his articles tended to conclude like so: separate the Hunnic question from the age-old question of Asiatic origins to study the history of Huns for itself. Perhaps, it was necessary: a scientific field is usually constructed by ruining the previous field, and it is certain that, in all respects, issues put forth by Maenchen-Helfen were a progress. However, to do this, he probably went too far. It seems possible today, with new Central Asian sources, not only to strengthen the political intuition of de Guignes—i.e., the Huns are well proclaimed heirs of the old empire Xiongnu—but to show that the Huns were part of the Xiongnu as well.
In his first article in 1945, Maenchen-Helfen continued Shiratori’s analysis, which had been little known to specialists in the Hunnic question, and showed that the main text, which was based on Chinese historiography to demonstrate the identity of the Xiongnu with Huns (Wei shu 102.2270), was interpolated from a gloss that totally distorted the meaning: the country of Su-tê, formerly called Yancai, subjugated by King Hu-ni of Xiongnu, was not the Crimean Sogdiana of Alans submitted by the Huns, as Hirth had previously thought, but Central Asian Sogdiana, deceptively close to the ancient name of the Alans in Chinese texts, Yancai. Thus, the reconciliation of Hun with Xiongnu saw itself deprived of any textual basis.1
However, in 1948, the Iranist W. Henning published a new reading of a text dated 313, a letter—the second of the Sogdian Ancient Letters—sent from the Gansu corridor to Samarkand by a Sogdian merchant named Nanaivandak. Henning showed that it was, without a doubt, possible that the plunderers Xwn of northern China described by this text, a name indistinguishable from the Huns of Western sources, were, in fact, the Xiongnu who put an end to the Jin Dynasty. This new document, which called the Xiongnu Xwn, took exactly the opposite view to Maenchen-Helfen’s thesis.2
This text is central to the issue at hand and it must be clarified that there is indeed no doubt that the Huns in Europe in the fourth and fifth centuries actually called themselves by this name. Not only all Western accounts agree in this sense, but this name is found unchanged in all populations affected by the extremely rapid Hunnic expansion, whether in the West, in the Balkans or in South Caucasus, a proof that it was Huns themselves who spread it. But, this is precisely the word used by the Sogdian merchant. Therefore, there are two possibilities:
— either Nanaivandak uses this word to describe properly the Xiongnu, that is to say Xwn is the transcript of the actual name of this people, which is Xiongnu in the Chinese transcription, and we must admit that the Huns have very well retained the name of the Xiongnu, in circumstances that remains to be explained, or
— our merchant uses a word from another source to be specified, in which case no link with Xiongnu can be established, and the only thing remaining is to explain why the Huns in Europe and a Sogdian merchant would use the same word.
It is true that Maenchen-Helfen warned repeatedly against such a reasoning: either one should reason based on names only, or only ethnographic and archaeological facts should be considered. But, presented so radically, this thesis is unacceptable. To quote the counter-examples he mentions many times, there is obviously a political link between Western and Eastern Romans, and, Charlemagne, like Diocletion and Joseph II, be that as it may, is a Roman emperor. The political dimension of a name can, by no means, be ignored; otherwise, much of the history of political ideas must be considered negligible. Byzantine Greeks proclaim themselves the heirs of the Romans, and are entitled to be so in terms of political history, and the Huns are equally able to proclaim themselves heirs of the Xiongnu. The steppe also has the right to have ideas, a political history, everything that conveys a proper name, and it is not clear why this would be irrelevant.
Aware of the major problem with this Sogdian testimony, Maenchen-Helfen published, in 1955, an initial response: he attempted to confront the evidence of the Second Ancient Letter with the oldest Greco-Latin texts, which also mentioned the Huns or similar names, but unrelated to the real Huns, to emphasize the purely nominal character of the testimonium sogdicum.3 Juxtaposing scholarly analyses on each of these items, Maenchen-Helfen’s reasoning does not stand up to the challenge: a first-hand evidence that calls the Xiongnu Huns, and comparison with purely literary references where reading Hun is not the result of an accident, is not likely to weaken. Only an internal explanation, giving an account of the use of the word Xwn by a Sogdian merchant in Gansu in the beginning of the fourth century, could lead to reject the link between Xiongnu and Huns.
This explanation was provided by H. Bailey. In 1954, in an article devoted to Harahuna of Indian sources, he set forth the hypothesis that the Sogdian Xwn could have been taken from the name of a hostile people cited in the Avesta, the Hyaona-: “Had the old name of the Hyaona survived among the Sogdians, which they then used for the Hsiung-nu? Or was this Sogdian Xwn taken directly from the name of the Hsiung-nu?”4 In 1959, Maenchen-Helfen resumed his account of this idea, while continuing to emphasize that the key was elsewhere, and lay not in the identity of names but in ethnological parallels.
Bailey’s proposal cannot be reduced to Maenchen-Helfen’s idea. But, the latter is what really interests us. Indeed, the purpose of Bailey was primarily to prove, at least partially, the Iranian character of the Xiongnu—which would have derived their name directly from the Hyaona—that was followed up by no one.
Maenchen-Helfen was no longer interested then in the Ancient Sogdian letters, which are not even mentioned in his World of the Huns, published posthumously in 1973. Pursuing his reasoning to its logical conclusion and having clearly stressed enough that what mattered was the material reality, he goes so far as to acknowledge the possibility of a vague phonetic link between the names of Xiongnu and Huns.
2. The Xwn of the Second Ancient Letter in recent historiography
The debate goes on, usually based on the ideas of Maenchen-Helfen, and the Second Ancient Letter continues to pose a problem in this context. So, when Sinor wants to rebut Henning’s reasoning, he writes: «The flaw in this argument is its disregard of the fact that the name Hun has been used consistently as a generic for many barbarian or barbarous peoples—for instance, in Byzantine sources in which Hungarians or Ottomans are often called Huns.»5 This presentation of the facts, which is a part of the reasoning of Maenchen-Helfen in 1955, is based on a strange conception of history: using examples after the Great Migration, Sinor does not run a great risk of being denied, because it is actually quite clear that the word Hun has already become generic. The whole problem is to clarify when. To be convincing, Sinor should have invoked the generic uses of the term Hun prior to the fourth century.
Daffinà also dealt with the text of the Second Ancient Letter, and placed it in the alternative explanation suggested by Bailey: Xwn is either the Sogdian form of Hyaona used generically to refer to the nomads, or it is a transcription of the name of the Xiongnu. To refute the second possibility, he evokes the differences between the actual pronunciation of the Chinese bisyllabic form “Xiongnu” and the way the Sogdian merchant heard it, i.e., Xwn. Xwn would merely be a distorted form of Xiongnu by an intermediate dialect, and hence a match with the name of the Huns in the West would be due to phonological happenstance alone.6 However, Xwn is by no means, in this case, a transcript of a Chinese form, local or not, but the name by which the Xiongnu called themselves. The Sogdians had direct contacts with the Xiongnu, completely independent of the Chinese, and transcribed the original name.7 That the Chinese form (Xiongnu = howling slaves) is of complex and uncertain shape, burdened as much by the phonology as by the games the Chinese played with its meaning in transcribing it, is certain; however, this does not take away any value from the completely independent Xwn transcription, even if it is one.
On the other hand, Daffinà further points out that the first possibility—that Xwn is a derivative of Hyaona—rests entirely on its only mention in the Second Ancient Letter, which limits its value. Sinor refuses Bailey’s reasoning, too.8
Yet, Bailey’s idea is the only one to save the currently dominant thesis in historiography—the absence of links between Xiongnu and Huns—and the evidence of the Second Ancient Letter. If we, thus, put aside the idea of pure chance, Daffinà’s position of which he himself recognizes its paradoxical nature, then only Bailey explains in a consistent manner why Nanaivandak could be misleading, because, if Nanaivandak did not misguide us, then the Huns of Europe had been using the name of the Xiongnu.
3. Hun, a generic pan-Iranian name?
When reduced to its simplest expression, that Sogdian Xwn comes from Avesta Hyaona, Bailey’s idea is appealing. It has recently been taken up and developed by S. Parlato, whose work is currently the only comprehensive attempt that could account for the use of the term Xwn by the Sogdians regarding Xiongnu.9 Parlato regards Hyaona as an adjective rather than a proper noun, designating a front of resistance outside of Iran against the attempts of centralization, and not a specific population. The word had a literary and epic character, and was spread in the steppe by bards of the Parthian Empire. It would have experienced an extraordinary success in the Scytho-Sarmatian world, and would have served as a generic name there to designate the hostile and demonic nomads.
This thesis is essentially of substituting in, by employing a simple word, a hypothetical expansion of a single tribal group through the Eurasian steppe and the Central Asia, for the nomadic Scytho-Sarmatian tribes traveled earlier but on the same space. It was not the Huns that would have moved; it would be a generic name that crossed the steppe space unified by the Iranian-speaking culture.
This thesis has several strengths: Besides Xwn as utilized by Nanaivandak, it also gives an account of the Hyōn of Sasanian sources, perhaps of the Indian Hūõa, and especially of Xοῦνοι mentioned by Ptolemy near the Don two centuries before any mention of the Huns. Nonetheless, I consider it false.
Philologically, there is hardly anything to say. Not having presented his ideas well, Bailey merely raised, never argued, a possibility of a link between Hyaona and Xwn. In fact, apart from Xwn, there is no known Sogdian word derived from one of the few words in Avestan with initial hy- that can serve as a point of comparison.10 The initial Avestan h disappears in Sogdian, and hence the derived Sogdian word should not start with an X. On the other hand, the Avestan -hi- or -hu- within the words also disappears in Sogdian, even though there are exceptions. And for lack of parallel case for the hy- initial, one cannot be certain of the normal derivation of Hyaona, and hence refute Bailey on this basis. Although it is quite unlikely, one cannot totally exclude philologically that Xwn is derived from Hyaona.
In terms of Sogdian religious history, there is equally little to hope: the known Zoroastrian fragments in Sogdian amount to a few lines, even though we now possess the painting of a big sacred codex, probably the Avesta, carried in procession in the streets of Pendjikent.11 It is impossible to know the role that the figure of the Avestan Hyaona might have played in the Sogdian religion.
The criticism must be historical. It has to deal with the following:
1) On the importance of Hyaona: the Hyaona are certainly important enemies in the Avesta, but, by no means, they play the leading part. Why then would they choose their name as a pan-Iranian word to designate hostile nomads? Is that role not already taken by the Tuirya? If it is conceivable that in a state as deeply Zoroastrian as the Sasanid Iran one is able to revive the old term of Hyaona in the form Hyōn to designate the Chionites in the Pehlevi sources, this religious context is entirely missing elsewhere, in particular in the steppe where there is no evidence of Zoroastrianism.
2) On the mode of dissemination, if it is epic and has no direct religious value: there is no evidence of any role of the Parthian minstrels in the steppe and to suppose that they have any influence on a tribe established in the North of the Balkans or having spread to Khotan to adopt a (hypothetical!) proper name derived from Hyaona.
3) Especially on the generic character: if the term Hun was generic, then how come the descendants of the nomadic Scytho-Sarmatian peoples did not use it (Where are the Ossetian Huns?), and more specifically, how is it that in the rare uses of the term prior to the fifth century, the term is never precisely generic? Finally, if really the word Hun was generic and of Iranian origin, then how is it that the Huns of Europe apply it to themselves?
On this last point, Sandra Parlato lightly claims: in response to the objection, she characterizes Huns of Europe as the Iranian-speaking Scythians who remained in the steppes of Southern Russia without attracting any attention before pushing the Goths away. However, one thing is perfectly clear: the Huns are not Iranian-speaking, as Maenchen-Helfen demonstrated a long time ago.12 Their personal names should then be explained in Iranian, which is absolutely not the case. They, therefore, had no reason to use a generic Iranian term, and even less to apply it to themselves. Moreover, it is unclear why these allegedly Iranian-speaking people would describe themselves by a term supposed to be so general, also not so accurate, and most importantly with strongly negative connotation. This point alone is sufficient, in my opinion, to show that the assumption of a generic Iranian name does not account for the available data, and adds very little to the proposed mode of dissemination.
Let’s go further on the supposedly generic character of the word by starting with a paradox: not only Hun is generic, but it is in Sogdian as well as in Khotanese, too. In other words, the recent developments in research seem to reinforce the thesis of Bailey and Parlato, and invalidate Daffinà’s reserve: Xwn is attested in numerous Sogdian texts. In addition to the Second Ancient Letter, one can quote, in chronological order:
— An unpublished fragment of a letter from the Tarim Basin like the Second Ancient Letter (and probably, judging by the paleography, of similar date): Xwn is located in a context impossible to clarify.
— 16 entries in the onomastics of the Sogdian graffiti in the Upper Indus, undated but assumed to be of the period between 3rd and 5th centuries (for grammatical and paleographic reasons, and because of the absence of Turkish names): I will treat them below.
— 7 in the documents from the mount Mugh of the first quarter of the eighth century.
In this last set, Xwn is clearly generic: It is either used to designate the Turks, nomads of the North to the East of the Sogdians, or is onomastic.13 Similarly, Bailey noted four Khotanese texts where Huna were mentioned: at least three of them use it in a generic sense.14 The fourth is more ambiguous but seems to belong to the context of 7th century, and it is, therefore, also generic.
However, all unmistakably generic references to the term Hun are later than the 4th century. It is of no use to show the generic nature of the word before this date, since it could be linked to the expansion of the Huns in Central Asia and Europe. In other words, to show that Nanaivandak baptized the Xiongnu by a generic name without any link to the actual name of these nomads, one can use only references prior to the Migration Period (Völkerwanderung).
4. References to the Huns before the Migration Period: the testimony of Zhu Fahu.
The references to the term certainly dated before the Great Migration are as follows:
— In Ptolemy, in the description of the European Sarmatians, a mention of a small tribe of Χοῦνοι with Roxolani on the Don;
— In two Indian Buddhist texts, translated into Chinese at the end of the third century and at the beginning of the fourth century, translations where Håõa is rendered by Xiongnu;
— Second Ancient Letter.
The Ptolemaic mention can hardly be of use to the argument: Completely unknown otherwise, this tribe, if indeed they were neighbors to Roxalani, a subgroup of Alans, could as well have taken as its name a generic term that is associated with Xiongnu in Central Asia, from where Alans emigrated, thereby borrowing a very prestigious name in an attempt to enhance themselves, as is often the case in the steppe.15 No data permits one to decide one way or another, and the possibility of chance cannot be excluded. The methodological error resulted from this reference, a source certainly familiar to Westerners, but in this case unusable.
The Indian references and their translation into Chinese are more interesting. The Tathagataguhya-sutra and the Lalitavistara both mention Hūõa in their lists, the former in one of the languages in which the Tathagata teach to the people of various countries and races, and the latter, the biography of Buddha, in the list of entries that the child Buddha teaches to his master. These texts predate the fourth century, and even though the former was lost, there is a Tibetan version, which provides the Hu-na lesson in the passage that interests us, while the latter is preserved. Both were translated into Chinese by Zhu Fahu, a monk native of Dunhuang, the first in 280 and the second in 308.16 In both cases, Zhu Fahu rendered Hūõa as Xiongnu.
These references suggest several areas of research:
1. What does the term Hūõa mean in Buddhist texts from the beginning of our era?
2. Is it a generic name in these texts?
3. Why did Zhu Fahu translate it as Xiongnu?
Hūõa does not belong to the Indian vocabulary and these two references are the oldest known. The term next reappears in the inscriptions to designate the invaders from the northwest plaguing India in the fifth century or in literary texts (Mahabharata).
The text of Lalitavistara, as such, is of little use, as the list of entries borrows from fantasy. At most, we can note that Hūõa were neighbors with the Chinese. It is quite different for the case of Tathagataguhya-sutra. The original text is lost, but there is an agreement that the Tibetan versions are extremely faithful to the original. Now, the Tibetan version provides
«Ka ça (corr. Ça ka) Pa hu pa (corr. Pa hla va) Thogar Yamana Kam po ce Kha ça Hu na Rgya-yul Da ra ta ...»
followed by numerous semi-fantastic peoples of the Indian geography, which suggests an Indian original:
«Çaka, Pahlava, Tukhara, Yavana, Kamboja, Khaça, Hūõa, Cïna, Darada, ...»
The Chinese translation of Tathagataguhya-sutra, in turn, gives the following:
“The race of Shi (Çâkya for Çaka), Anxi, Yuezhi, Daqin, Jianfu, Raodong, Qiuci, Yutian, Shale, Shanshan, Wuqi and the early and later kingdoms (the Turfan region), Xiongnu, Xianbei, Wu, Shu, Qin, the Mo, the Yi, the Di, the Taluoduo, the ignorant, the savages, and also...”17
The structure of the Indian list is absolutely clear: the beginning of the text constitutes a panorama of the foreign peoples as the Indian editors could conceive them. These people are in contrast to the imaginary peoples or those stemming from a distant past and half-legends, which are rejected in the end. Hūõa had no reason to be placed at the beginning of the list if they were not a people having a geographical or ethnic reality of the same order as that of the Parthians, the Bactrian, the Greeks, the Cambodian of the Hindu Kush, or of the Khasa of the Himalayas. This term has a well-defined geographical and cultural meaning in the eyes of the Indians, though in a large scale. They are, on the other hand, well-documented in the Sanskrit literature, as well as in the saga of the religious codes or texts.
Yet, the Hūõa are totally absent from the earlier literature, and their presence in this list cannot be explained as part of the divisions of the world specific to the Indian culture. They are, indeed, a part in this text of big ethnic or geographical divisions, as they were named at the time. This list is rather precisely datable: it corresponds to the context of the 1st century before our era or the beginning of our era, and co-exist with Parthians, Greeks and Sakas, before the conquest of India by Kushans. Among the political powers of the time, besides the Chinese, are, thus, mentioned Hūõa. In other words, regardless of the Chinese translation of Hūõa, this leads to Xiongnu, the empire of shanyü, the only major Asian empire whose mention makes sense chronologically and whose name can be approximately closer to Hūõa, to be the most likely candidate already: among the languages which the Tathàgata teach people is that of the Xiongnu alongside the Chinese, Greeks, Parthians.
The translation of Hūõa for Xiongnu is to reinforce this finding naturally. This is not a Chinese- fashion interpretation, vague and generic: all the names in the Indian text making sense in a framework of the Chinese thought are rendered word-by-word in translation or transcription, and possibly clarified by a political update corresponding to the context in 280. Only the things purely related to the Indian way of things are deleted (Khasa) and replaced with others. Now, just like the Pahlava are indeed the Parthian Arsacid Empire (Anxi), the Tukhàra the Yuezhi, the Yavana the Greeks (Daqin, roughly the East Roman Empire, in other words, the Hellenistic world), and China the Qin, nothing allows one to doubt in this text that the Hūõa are precisely the Xiongnu. Zhu Fahu does not consider Hūõa as a generic name: he had every opportunity to make them appear further away, in the quasi-historical part of a list, which he did not hesitate to revise in depth until it no longer made any sense to him, with Mo, Yi, and Di, peoples of the far North in Chinese antiquity, corresponding much better to the idea of a generic name for “barbarian peoples,” and add them to the Indian text. He could also simply have transcribed it, as in the name of Dardes, or even eliminated it and replaced it with others, as in the case of Khasa. He does none of this, but, as in Lalitavistara, returns Hūõa for Xiongnu.
He even goes beyond that: he updates the term in a manner that fully confirms the precise political sense of Hūõa in the original text. On closer inspection, there are, in fact, two Chinese names that correspond to the Hūõa in the Tathagataguhya-sutrà: Xiongnu and Xianbei. However, in 280, Xianbei occupy exactly the place which Xiongnu occupied at the time of the drafting of the original text.18 These are the main nomadic enemies in steppes North of China, with the Xiongnu grouped further South in Shansi. In other words, with Zhu Fahu adding Xianbei to Xiongnu, the normal translation of Hūõa (as evidenced, let us recall, in Lalitavistara by the same translator) annotates the term of Hūõa politically: Xianbei structurally plays the political role which was that of Xiongnu before. He proceeds elsewhere with the following term similarly: Cãna is given as “Wu, Shu, Qin territory.” Wu and Shu are both the names of kingdoms in South China (lower Yangzi and Sichuan) in the 3rd-century China, the last of which, Wu, falls exactly in 280, while Qin returns to Cãna at the same time, but, no doubt, here also means northern China. In other words, the generic Hūõa has nothing to do specifically with northern China.
All in all, the use of the term Hūõa in these texts has a political sense, that of Xiongnu and goes back up to the period when they were the main nomadic adversaries of China and principalities of Central Asia. The Indian editors quite logically included them in their lists and Zhu Fahu normally rendered the term as Xiongnu.
Until now, we were not interested in Zhu Fahu much: yet, he was very well known, since he was Dharmarakṣa, the principal introducer of Buddhism in China in the third century.19 In spite of the name of Zhu, which he took from his master, Zhu Fahu was from a Bactrian family settled in Dunhuang for several generations, a true representative of this milieu of large merchants with which Buddhism was established in China. He was Sinicized but remained deeply tied to the Central Asia where he traveled on several occasions, and which he knew “all” its languages. He maintained, in particular, his links with Gansu, and, after his installation in central China, returned regularly to Dunhuang. In other words, Nanaivandak and Zhu Fahu did not belong to radically different circles, although a generation separated them and even though Zhu Fahu was far better integrated into the Chinese environment. Both were Iranian-speaking émigrés, and both could, according to the hypothesis of Bailey and Parlato, understand Xwn and Hūõa in a generic sense: however, that was not the case. Zhu Fahu used it with precision and his wisdom fully supported the most immediate interpretation of the testimonium sogdicum. For both, the Huns were indeed the Xiongnu.
The reasoning could end here: the only alternative hypothesis proposed to account for the use of Xwn by Nanaivandak about the Xiongnu in northern China does not stand up to scrutiny. It gives an account neither of the name of Huns of Europe nor of the Hūõa term of the Indian texts, and is not based on any mode of diffusion identified. The Indian authors and Zhu Fahu renders Xiongnu as Huns and vice versa in contexts that exclude it as a generic name. Nanaivandak is right to name Xwn the Xiongnu. The Huns arriving in Europe from 370 very well used the name of the Xiongnu, which is not, contrary to what Maenchen-Helfen wrote, devoid of any meaning. In the current state of the data, the burden of proof rests entirely on those who want to separate radically the Huns of Europe from Xiongnu.
To proceed further, it is necessary to be interested in the Central Asia. These Huns who appeared on the Volga in 370 with an old imperial name that originated from the Far East inevitably came from the East: this was also indicated by all the Graeco-Latin sources, and Maenchen-Helfen was very misguided to minimize this fact.20 Proponents of an absolute distinction between the Huns and the Xiongnu are, moreover, careful not to explain how it was possible that the Huns, supposedly Pontic, were in Central Asia. However, the knowledge on the history of Central Asia have been profoundly reshaped since the time of Maenchen-Helfen and Bailey. Much of the textual arguments or numismatics employed half-a-century ago for the history of Huns in this area cannot be accepted anymore today. In addition, our archaeological knowledge is infinitely better. I would, therefore, like to present this new information and propose some hypotheses.
II. New data on the Huns in Central Asia (4th and 5th centuries)
1. Numismatic data
First of all, the numismatics. The reading oiono on Ephthalites currency had confirmed the Byzantine texts speaking about the Huns in Central Asia and was one of the most frequently quoted in debates on Iranian Huns: it is now completely abandoned.21
The reading αλχοννο on the coins from the late fourth century Kapisa and Gandhara, struck with re-used corners of Shapur II, is now generally corrected as αλχαννο by establishing the link with the Indian legend rajalakhana (i.e., rajah alakhana), found on one of the issues,22 which would not make it possible any more to establish the link with Alxon and Valxon of the Armenian Geography.23
Another reading now disputed is that of βαγο κιδαρο on the dinars of the treasure of Tepe Maranjan close to Kabul and buried under Shapur III (383–388): these currencies clearly carry only κι followed by a highly variable number of undifferentiated o. They could not provide the basis for a timeline whereas all the other sources agree to place the Kidarite Huns (Oὄννων τῶν Κιδαριτῶν) in the period 420–440, and not in the 4th century.24 Still on the subject of Kidarites, one reads from now on kyδr on certain currencies from Samarkand reflecting an expansion of their empire in Sogdiana.25
This triple change has an important consequence: without oiono, and reading αλχαννο, we no longer know how the invaders ravaging Central Asia called themselves. There are no more Huns on currencies. Moreover, the text of Ammianus on Chionites and that of Faustus of Byzantium in the 370s become our only sources on the events of the fourth century.
2. Archaelogical data
Two sets of archaeological data emerged to renew the question profoundly. Maenchen-Helfen had had time to become acquainted partially with the first, but the second, closely related to the Soviet work in Central Asia and depending on the question of the date of Kanishka, has been considered relatively recently.
In his book published posthumously in 1973, Maenchen-Helfen recognized, after having denied it a long time, that the Hunnic cauldrons had their origins in Xiongnu cauldrons. Furthermore, the sites of these objects, buried close to springs or rivers, were similar in Inner Asia and Hungary, reflecting not only material but also ritual and cultural continuity.26 In this posthumous and unfinished work, he did not draw any conclusion. The knowledge on these cauldrons has since progressed. It emerged not only that the Hunnic cauldrons of Pannonia arose from the Xiongnu cauldrons, but that one could follow a typological evolution of the simple shapes of Ordos to the European complex forms. In addition, roads that must be necessarily considered as migration routes due to the continuity of worship now appear clearly.27 One of them is extremely northern and traverses the southern Siberia from East to West up to the gap of Yekaterinburg in the Urals. On its route, the bulk of the finds of Xiongnu cauldrons are concentrated between the Mongolian steppe and the Hungarian plain. Yet, by applying precisely the same typological rules, I am not sure that a direct link is necessarily established between the Xiongnu groups that appear on the Upper Kama at the mouth of the Ural and the Huns who invaded the Western world. There are several models of Xiongnu cauldrons quite clearly differentiated by the shape of their handles. Cauldrons found on the northern route to the Upper Kama belong to a very specific model never found in Hungary. On the other hand, cauldrons found along a second route, much farther south, correspond exactly to the findings of the Hunnic world. They are far fewer, but from Urumqi and Altai to Central Asia, to the Volga River and finally to the Hungary they trace what seems to be the story of a migration, which by its journey would give an account of the whole of the available data.
Other data set concerns sedentary Central Asia. The Soviet work was extremely important both in Uzbekistan and Tadjikistan, as in Afghanistan. The entire chronology of Central Asian sites is highly dependent on the monetary series inaugurated by the Kushan emperors, continued by Kushano-Sassanians and their barbaric followers. However, the chronology of the Kushan Empire remained disputed for a long time, and the Soviet school28 was characterized by an extreme position, dating the first year of Kanishka to 371, while the emerging consensus is now 127–8. Once the chronology of the sites was reconsidered, the picture becomes rather clear and sets the North and South of Amu Darya, i.e., Sogdiana and Bactria. In Sogdiana and Bactria, the 4th century is marked by a profound change in the population,29 a regression of the sites, layers of fire and the abandonment of certain irrigation networks. Further North along the Syr Darya, the irrigated agriculture is also permanently reached, and it seems, from the ceramics, that a part of the population took refuge in Sogdiana. It is, in particular, the case of the Džety-Asar culture, in the delta of Syr Darya, that was largely decimated at the end of the 3rd century or in the 4th century (multiple destroyed sites, levels of fires). A part of the population arrived in the Caucasus while another moved southward to Fergana.30 The inhabitants of the Kaunchi culture (along the middle course of Syr Darya) took refuge in Sogdiana.31 Finally, in Khwarezm, the 4th century clearly marks a break. However, in the 5th century, the situation is different: while Bactria continues to bear the scars of the fighting, the Sogdiana enters a phase of very fast rebuilding, undoubtedly partly thanks to the refugee populations of Syr Darya and Bactria, as well as an agricultural urban viewpoint. Political stability seems assured in the long term, because canals are re-grooved, and especially a planned urban development is implemented across the Valley of the Zerafshan.32 The 5th century sets the stage for the long economic dominance of Sogdiana over Central Asia in early Middle Ages.33 Now, Weishu 102.2270 provides the ethnic identity of the dynasty that, for at least three generations in the fifth century, rules Sogdiana and presides over this development: they are Xiongnu.
3. Textual data
From a textual point of view, the appearance of numerous Bactrian documents essentially clarified the onomastics. Thus, the name of Grumbates, the king of the Chionites, ally of Shapur II who took part in the siege of Amida (Diyarbakir) in 359,34 is now affirmed to be of a prince, shortly before 470, of the kingdom of Rob (North of the Hindu-Kush) in the form Gorambad γοραμβαδο.35 The date of this document, subsequent to the statement of Ammianus Marcellinus, does not make it possible to decide if this name is authentically Bactrian, in which case Chionites would certainly be also, or if it were simply integrated into the local noble onomastic stock. The documents and the Bactrian seals also give the local form of the name of the Ephthalites, ηβοδαλο, and the personal names of their kings Khingila εþκιγγιλο and Toramāõa τωρομανο.36 The Bactrian documents also lead to give up certain identifications: thus, the Himalayan polyandry used by Enoki to assign a mountain and Pamir origin to Ephthalites is now attested in Bactria a long time before their arrival: the Chinese records, which mention it, confuse, under the name of Ephthalites, the wandering elites and the local populations which they dominate. Many texts have not yet been published, and we can reasonably expect that they will provide information about peoples present in Bactria in the fourth century as they already do for later periods.
On the Sogdian side, the appearance, reported above, of the name Xwn within Sogdian onomastics in the caravaneers’ graffiti of the Upper Indus is interesting: even if they do not have the same political significance as the now-abandoned readings of the currencies, they ensure, nevertheless, the presence of Huns in contact with the Sogdians sometime between the 3rd and 5th centuries. The name is mentioned 16 times, and is in the third position by the number in the onomastic corpus of these graffiti: this remarkable integration to the stock of the Sogdian names reflects a period of merger of populations, not of hostile relations.37 Yet, it is necessary to specify the date of these inscriptions. Now, a detail makes it possible to think that these graffiti mentioning of Xwn are among the last engraved. Indeed, contrary to the other names, Xwn is completely absent from patronyms. There are 16 Xwn son of X, and no X son of Xwn. These Xwn have fathers with perfectly Sogdian names. The Xwn name is only in this context, which leaves little to chance.38 On the other hand, a number of reasons suggests that the last graffiti were registered in the 5th century:39 Those mentioning Xwn would then be a part of them, and it is in the 5th century that we must place the fusion of the populations. Their appearance thus confirms independently the Chinese and Byzantine texts on the presence of Huns in Sogdiana: the Sogdian dynamism of the 5th century of our era under a Kidarite dynasty, qualified as Xiongnu by the Chinese sources and as Hunnic by the Byzantine sources, forms indeed an ideal context for the Xwn name to appear in the Sogdian onomastics. In my opinion, these Sogdian Xwn were born under the Sogdian Kidarite dynasty, and engraved their names on the rocks of the Upper Indus before 460 and the sudden Ephthalite expansion in Bactria—a region that separates Sogdiana from these passes—blocked the road.
Other texts—Chinese, Armenian, Syriac, Byzantine and Indian—have long been known. Hardly any attention has been paid, however, to the various reconstructions suggested in the fact that the Chinese sources mention, at the beginning of the 5th century, the “remnants of the descendants of the Xiongnu,” far to the Northwest of the Ruanruan, i.e., around the Altai (Weishu 103.2290). There is no reason to dispute the quality of the information: Weishu is very parsimonious in its use of the term Xiongnu,40 and these Xiongnu are the only ones among a list of the neighbors of the Ruanruan. However, this information is very important: by itself, it knocks down one of the main arguments used against the hypothesis of a migration of the Huns from the former Xiongnu territory, the absence of any mention of the Xiongnu to the North past the 2nd century. They actually survived far North, completely beyond the field of view of the Chinese sources. That they did not form an empire and that they were reduced to only being the heirs of the former Xiongnu matter little: the tribal political identity was preserved.
Another indication, in Weishu 102.2278–9, although known, has not received all the attention it deserves: Ephthalites are said to have migrated from Altai to the Central Asia around 360–370:
“Country of the Yeda, of the race of the Great Yuezhi, they are also said to be another branch of the Gaoju. They originally come from a region to the North of Sai. Having left the Altai southward, they settled West of Khotan. Their capital is more than 200 li South of the Oxus, and 1100 li to Chang’an.”
What adds the Tongdian, based on the original text of the Weishu, which it often preserves, is that the departure from the Altai took place 80 to 90 years before the reign of Wen Cheng Di of the Later Wei (452–466).41
4. A proposal for a reconstruction
In light of the above data, it is time to propose a reconstruction with the usual precautions: it is a matter of giving an account, in the simplest manner, of the greatest possible number of data, leading to the only likely result.
In my opinion, a group of tribes, gathered under the political identity of Huns, left the foothills of Altai in several waves between 350 and 360, some preyed on the Syr Darya, the others on the Volga.
Several facts indicate that the invaders of Europe as well as Central Asia relate ultimately to the Altai: it is the Altai that has the largest concentration of advanced Hunnic cauldrons in the East, and the only direct prototype of the Hunnic cauldrons of Hungary; it is from there that the Ephthalites have undoubtedly departed; this is the area that the Chinese sources still located the “remnants of the descendants of the Xiongnu” at the beginning of the 5th century. Finally, the devastation in Central Asia is from the North and East: the populations of Syr Darya took refuge in the direction of the South or of the Caucasus.42
The timeline is equally consistent: the diffusion of the ceramics of Syr Darya in Sogdiana shows that the devastation in Central Asia form an integrated whole, and not a succession of isolated episodes; from 350 on, Sassanians face on their northeastern border non-Iranian adversaries (the practice of cremation) whom they name Chionites; Eighty to ninety years prior to 452, or around 360–370, tribes, including the Ephthalites, leave the Altai in the direction of the South and of Central Asia; in 370 at the latest, Hunnic tribes appear between the Volga and the Don.43
Finally, the Hunnic identity of these tribes is in little doubt; given the name they bear, in Europe as well as in the texts in Central Asia, the possibility of a generic name is ruled out: still, it is necessary to agree on the meaning of this identity. The Chinese source makes it possible, on the one hand, to affirm that, near the Altai, nomadic groups have continued to proclaim themselves the political heirs of the Northern Xiongnu far beyond the field of view of the Chinese sources between the 2nd and 4th centuries. The concentration of Xiongnu cauldrons in the area and the continuity of their use in rituals show, in addition, that this heritage is also partly religious. These Altaic nomads could come down to tiny scattered Xiongnu bands, or in contrast form large groups; nevertheless, it is this political and cultural Xiongnu identity that helped unite the tribes. In the Steppe, the language and blood are irrelevant.
Alram, M., “Alchon und Nezak. Zur Geschichte der iranischen Hunnen in Mittelasien,” in La Persia e l'Asia Centrale da Alessandro Magno al X secolo, (Atti dei Convegni Lincei, 127), Rome, 1996, pp. 519–554.
Alram, M., “A Rare Hunnish Coin Type,” Silk Road Art and Archaeology, 8, 2002, pp. 149–153.
Ammianus Marcellinus, History, various eds., translation and commentary, Paris: Les Belles Lettres, 1968–1977.
Bailey, H. W., “Harahuna” in Asiatica. Festschrift Friedrich Weller. Zum 65. Geburtstag gewidmet, Leipzig, 1954, pp. 12–21.
Bailey, H. W., Indo-Scythian Studies being Khotanese texts, VII, Cambridge University Press, 1985.
Burjakov, Ju. F., “À propos de l'histoire de la culture de la région de Tashkent au 1er millénaire avant notre ère et au 1er millénaire de notre ère,” in P. Bernard, F. Grenet, eds., Histoire et cultes de l'Asie centrale préislamique: Sources écrites et documents archéologiques, Paris: CNRS, 1991, pp. 197–204.
Chavannes, É., “Les Pays d'Occident d'après le Wei Lio,” T'oung Pao, series II vol. VI, 1905, pp. 519–571.
Daffinà, P., “Chih-chih Shan-Yü,” Rivista degli Studi Orientali, 1969, XLIV–3, pp. 199–232.
Daffinà, P., “Stato presente e prospettive della questione unnica,” in S. Blason Scarel, ed., Attila. Flagellum dei?, Rome, 1994, pp. 5–17.
Enoki, K., “Sogdiana and the Hsiungnu,” Central Asiatic Journal, 1, 1955, pp. 43–62.
Erdy, M., “An Overview of the Xiongnu Type Cauldron Finds of Eurasia in Three Media, with Historical Observations,” in B. Genito (dir.), Archaeology of the Steppes, Naples, 1994 , pp. 379–438.
Grenet, F., “Les ‘Huns’ dans les documents sogdiens du mont Mugh (avec un appendice par N. Sims-Williams),” Études Irano-aryennes offertes à Gilbert Lazard, (Cahier de Studia Iranica, 7), 1989, pp. 165–184.
Grenet, F., “Crise et sortie de crise en Bactriane-Sogdiane aux IVe–Ve s. de n.è.: de l’héritage antique à l’adoption de modèles sassanides,” in La Persia e l'Asia Centrale da Alessandro al X secolo, (Atti dei Convegni Lincei, 127), Rome, 1996, pp. 367–390.
Grenet, F., “Regional Interaction in Central Asia and North-West India in the Kidarite and Ephthalite Period,” in N. Sims-Williams, ed., Indo-Iranian Languages and Peoples, (Proceedings of the British Academy 116), London, 2002, pp. 203–224.
Henning, W. B., “The Date of the Sogdian Ancient Letters,” Bulletin of the School of Oriental and African Studies, XII 3/4, 1948, pp. 601–615.
Hewsen, R. H., (introduction, translation, and commentary), The Geography of Ananias of Širak (AŠXARHAC’OYC’). The Long and the Short Recensions, (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe B (Geisteswissenschaften), 77), Wiesbaden, 1992, 467 p.
de la Vaissière, E., Histoire des marchands sogdiens, (Mémoires de l’IHEC, 32), Paris, 2002.
de la Vaissière, E., Riboud, P., “Les livres des Sogdiens,” Studia Iranica, 32–1, 2003, pp. 127–136.
Levina, Larisa M., Ėtnokul'turnaja istorija Vostočnogo Priaral'ja. I tysjačeletie do n. ė.—I tysjačeletie n. ė., Moscow:Izdat. Firma Vostocnaja Literatura, 1996, 396 p.
Lévy, S., “Notes chinoises sur l’Inde. V Quelques documents sur le bouddhisme indien dans l'Asie centrale,” Bulletin de l'École Française d'Extrême-Orient, V, 1905, p. 253–305.
Maenchen-Helfen, O., “Huns and Hsiung-nu,” Byzantion, 17 (1944–45), 1945, pp. 222–243.
Maenchen-Helfen, O., “Pseudo-Huns,” Central Asiatic Journal, 1, 1955, pp. 101–6.
Maenchen-Helfen, O., “Names of the Hsiung-nu,” Central Asiatic Journal, 6, 1961, pp. 249–261.
Maenchen-Helfen, O., The World of the Huns. Studies in Their History and Culture, Berkeley, 1973.
Marshak, B., Raspopova, V., “Les nomades et la Sogdiane,” in H.-P. Francfort, ed., Nomades et sédentaires en Asie centrale. Apports de l’archéologie et de l’ethnologie, Paris:CNRS, 1990, pp. 179–185.
Parlato, S., “Successo euroasiatico dell’etnico ‘Unni’,” in La Persia e l’Asia Centrale da Alessandro al X secolo, (Atti dei Convegni Lincei, 127), Rome, 1996, pp. 555–66.
Procopius, History of the Wars: edited and translated by Dewing H. B., History of the Wars, I, Cambridge (Mass.):Harvard University Press, 1914, 583 p.
Sedov, A. V., Kobadian na proroge rannego srednevekov’ja (Kobadian at the Beginning of the Early Middle Ages), Moscow:Nauka, 1987, 199 p.
Semënov, G. L., “Sogdijskij gorod v rannem srednevekov’e: formirovanie plana” (Sogdian Cities of the Early Middle Ages: Formation of the Plans), in G. Smirnova, ltogi rabot arxeologičeskix ekspedicij Gosudarstvennogo Ermitaža, Leningrad, 1989, pp. 128–140.
Shiratori, K., “A Study of Su-t’ê, or Sogdiana,” Memoirs of the Research Department of the Toyo Bunko, 2, 1928, pp. 81–145.
Sims-Williams, N., Sogdian and other lranian Inscriptions of the Upper Indus, I-II, (Corpus Inscriptionum Iranicarum, II/III), London, 1989–1992.
Sims-Williams, N., New Light on ancient Afghanistan. The Decipherment of Bactrian, London, 1997.
Sims-Williams, N., “Ancient Afghanistan and its Invaders: Linguistic evidence from the Bactrian documents and inscriptions,” in N. Sims-Williams, ed., Indo-Iranian Languages and Peoples, (Proceedings of the British Academy, 116), London, 2002, pp. 225–242.
Sinor, D., “The Hun Period,” in D. Sinor, ed., Cambridge History of Early Inner Asia, Cambridge, 1990, pp. 177–205.
Tongdian, by Du You, Changsha, Yuelu shushe, 1995.
Wei Shu, by Wei Shou, Beijing: Zhonghua shuju, 1974.
Yü, Ying-shih, “The Hsiung-nu,” in D. Sinor, ed., Cambridge History of Early Inner Asia, Cambridge, 1990, pp. 118–149.
Zeimal’, E. V., “The Political History of Transoxiana,” Chapter 6, The Cambridge History of Iran, 3/1: The Seleucid, Parthian and Sasanian Periods, Cambridge University Press, 1983, pp. 232–262.
Zürcher, E., The Buddhist Conquest of China, (Sinica Leidensia, XI), 2 vols., 2nd ed., Leiden:Brill, 1972, 470 p.
1 Maenchen-Helfen, 1945, using Shiratori, 1928, p. 98 and following. Also on the same subject, see Enoki, 1955, p. 45 and following. Others before him had disproved a second argument, which made a minor event in the history of Xiongnu—the installation of one of their rulers in the first century b.c. north of today’s Tashkent—the source of the Xiongnu's westward migration; so the episode was short-lived: Daffinà, 1969.
7 One can possibly consider an intermediary, such as Yuezhi or Wusun, which does not affect its independence from the Chinese transcription.
8 Sinor sees in only Chionites, adversaries of Sassanids at the 4th century, the descendants of Hyaona: Ammianus Marcellinus who knows Chionites and Huns never establish the link between the two (Sinor, 1990, p. 179).
20 Maenchen-Helfen, 1973, pp. 444–7: the reasoning on Jordanes is particularly strange. Jordanes located the Huns far to the East, alongside the Seres, and Maenchen-Helfen hypothesizes that Jordanes borrowed this location, like many other data, from Cassiodorus, who himself would have taken it from the Periegesis of Dionysius. None of this is proven. But later, Maenchen-Helfen stresses that no manuscript of the Periegesis gives the name of the Huns and that contemporary editions depend on a correction by Müller, the editor. But if the name of the Huns was not there, how could Jordanes or Cassiodorus borrow it? Finally, the names of Seres and “Huns” are, by no means, close in Denys’s text, and Maenchen-Helfen supposes that Cassiodorus drastically reduced the list of the Eastern peoples of this text, with a result, therefore, of making Huns the neighbors of Seres entirely at random. But why Cassiodorus would have proceeded thus? In other words, faced with texts directly contrary to his theories, Maenchen-Helfen produces ad hoc even contradictory assumptions. Jordanes was not at a loss to talk about the Huns, and his testimony is corroborated by other, certainly more vague, texts including that of Ammianus Marcellinus (XXXI, 3, 1) who places them far to the Northeast.
23 Translated by Hewsen, 1992, p. 75. For my part, I am less certain of the validity of this correction: If a large number of emissions carry -a-, the fact remains that some of them, among the earliest, carry -o-. We could be witnessing a phonetic or graphic evolution. It would be highly surprising that the identity between the name of the Alxon located by the Armenian geographer on the Syr Darya and that of the new dynasty αλχοννο or αλχαννο in Gandhara is of pure chance, while it is known that a certain number of nomadic peoples passed the mountains to the South during this period.
28 Still, it would be necessary to distinguish between the school of Leningrad, dating Kanishka to 371 following the works of Zeymal’, and the school of Tashkent, much more cautious, and in favor of a date very close to the date now accepted.
37 Other sets of documents that allow to study Sogdian onomastics, documents of the Mugh Mountain from the 8th century, and Chinese documents from Turfan (7th–8th centuries) hardly mention this name.
38 There are, thus, to cite a few examples, 32 mentions of the name Nnyβntk including 10 in the position of a surname; similarly, there are 26 Pys’k including 10 patronymics, 12 cytβntk including 3 patronymics, 9 wxwšwβntk including 5 patronymics, 8 δwr’k including 3 patronymics, 8 m’ymrγc including 5 patronymics, 6 wxwšwγβ”r including 3 patronymics, 4 ‘βy’mnβntk including 3 patronymics... on the whole, all the names represented in more than 5 inscriptions (but also most of those that are present more than 3 times) are attested in the patronymic position, and only Xwn makes an exception.
39 The absence of Turkish first names leads to excluding the sixth century, while the presence of the given name and the surname m’ymrγc seems to exclude the 4th century, as the city of Maymurgh seems to flourish in the 5th century.
40 One counts in the text about forty occurrences of the term, mostly associated with the Southern Xiongnu or rhetorical comparisons with the Han period. References to the contemporary Xiongnu other than that in the South are extremely rare; there are only a total of three in Chapters 102 and 103 that are devoted to the countries of the West and North, i.e., geographically the entire former Xiongnu Empire. One is that cited above (103.2290), the second is in the famous passage on the conquest of Sogdiana by the Xiongnu (102.2270), and the third describes the fights with Ephthalites, the White Huns of the Byzantines. Procopius emphasizes very clearly the Hunnic identity of the Ephthalites: “At a later time, the Persian King Perozes became involved in a war concerning boundaries with the nation of the Ephthalite Huns, who are called White Huns, gathered an imposing army, and marched against them. The Ephthalites are of the stock of the Huns in fact as well as in name (Έφδαλἴται δὲ Οὐννικὸν ὲν ἔθνοζ ἐισί τε καὶ ὀνο άζονται); however, they do not mingle with any of the Huns known to us.” Procopius, History of the Wars, I, 3, 1–8 translated by Dewing vol. I p. 13–5.
42 Other reconstructions, based on the idea of a migration but with different starting points, face insurmountable difficulties. The Soviet school has long argued that the Huns were indeed the descendants of Xiongnu, but with Xiongnu having migrated to the southern Urals for a long time, where they merged with Ugric populations before heading down to the South in the fourth century (references in Maenchen-Helfen, 1973, p. 447 n. 21; see also: L. N. Gumilev, Xunnu, Moscow, 1960, chap. 15). This reconstruction avoids the pitfalls of the generic name and could even explain the Hunnic presence in Central Asia: one can possibly imagine a conquest of the Central Asia by Hunnic groups from the Northwest. But in this context, it is very difficult to understand why cauldrons of Hungary are not precisely of the type of the Urals but go back up to samples found in the Altai or Urumqi. To explain the presence of the lower Syr Darya ceramics on the Caucasus, count down from the supposed invasion, mainly to account for the Chinese text making the departure of the Ephthalites from the Altai to 360. To place the starting point halfway, in the Kazakh steppe, on the Syr Darya or the Semirechye does not solve anything: The Wei lue, like the archaeology, forbids one to see in these areas any Xiongnu presence in the 3rd century (see the translation in Chavannes, 1905). Note in particular that the Xiongnu miniature pots found in the Dzety-Asar excavations predate our era: see Levina, 1996, p. 188.
43 Let us recall that the distance hardly matters: even traveling only 6 months per year, at the rate of 2 km to the West per day, it takes only 5 years to cover the distance between the Altai and the Volga.