The Slavs and the Avars

Omeljan Pritsak












There is general agreement among western scholars that the Avars were instrumental in the appearance of Slavs on the stage of history [1]. As the Huns caused the Germanic peoples (along with the Iranian Alans) to migrate and to develop new political groups, so - a familiar thesis runs - it was the Avars who caused the Slavs to move and to develop. Yet the Avars have remained, so to speak, stepchildren in historical studies, and the raison d'être of their activities has hardly been explored [2].



(*) I should like to express my sincerest thanks to my friend Professor Horace G. Lunt for helpful critical remarks and substantial editorial aid.


(1) For example, J. Peisker, «The Expansion of the Slavs», Cambridge Medieval History, vol. 2, Cambridge 1913, pp. 418-458; Arnold J. Toynbee, A Study of History (Oxford Univ. Press), vol. 2 (1963), pp. 317-319, vol. 3 (1962), p. 22. See now Wolfgang Fritze, Frühzeit zwischen Ostsee und Donau, Berlin 1982, 47-99.


(2) For bibliography on the Avars, see: Gyula Moravcsik, Byzantinoturcica, vol. 1, 2nd ed. Berlin 1958, pp. 70-76; Denis Sinor, Introduction à l’étude de l’Eurasie Centrale, Wiesbaden 1963, pp. 231-232, 265-267; Arnulf Kollautz, Bibliographie der historischen und archäologischen Veröffentlichungen zur Awarenzeit Mitteleuropas und des Fernen Ostens, Klagenfurt 1965, 25 pp. (unreliable); Aleksander Avenarius, Die Awaren in Europa, Amsterdam-Bratislava 1974, pp. 269-283; and Wojciech Szymański, in W. Szymański and Elżbieta Dabrowska, Awarzy. Węgrzy, Wrocław 1979, pp. 123-138. Cf. Samuel Szádeczky-Kardoss, Ein Versuch zur Sammlung und chronologischen Anordnung der griechischen Quellen der Awarenzeit nebst einer Auswahl von anderssprachigen Quellen, Szeged 1972, and the improved Hungarian version, «Az avar történelem forrásai [Sources for Avar history]», Archaeologiai Értesitő 105 (1978) 78-90; 106 (1979) 94-111, 231-243; 107 (1980) 86-97, 201-213. See also the Appendix at the end of this paper.





Since World War II, archeological research on Avar sites, especially in Hungary and Slovakia, but also other Slavic countries, has developed rapidly. The state-sponsored academic institutions have been conducting excavations on a large scale according to well-prepared plans, and reports on the new materials are published almost immediately. Specialists have been sifting the old and new data collected from sites in territories once under Avar domination, in an effort to puzzle out the ethnic, economic, and social structure of the Avar realm. In this way it is the archeologists who have taken the study of the Avars into their hands [3]. The conjectures of the Slavic scholars usually overrate the roles of Slavs in Avar society [4], while the Hungarians tend to go to the other extreme. There are two reasons for this. The first is the paucity of written sources on the Avars. Although the Avars were a successful imperial elite, they apparently were, like their predecessors, illiterate [5]. All suceeding Eurasian rulers used some sort of writing system and recorded at least a few historical messages [6]. The Avars, however, left no writ-



(3) Archeological research to about 1954 is conveniently listed in the standard work by Desző Csalylány, Archäologische Denkmäler der Awarenzeit in Mitteleuropa. Schrifttum und Fundorte, Budapest 1956, 244 pp., 1 map. In my Appendix below I update his bibliography (pp. 17-63), but I repeat the pre-1954 items which are basic for the history of the Avars.


(4) See the sound criticism by the German archeologist Helmut Preidel in his «Awaren und Slawen», Südostforschungen 11, Munich 1952, 33-45.


(5) Bone artifacts with short (ritual?) runiform signs have been found in certain Avar-period graves in Hungary (e.g. Jánoshida, kom. Szolnok; Szentes-Felsőcsordajárás, kom. Csongrád). Although the provenance of the script has not yet been determined with any certainty, some scholars speak of «Avar runes»; cf. Jovan Kovačević, Avarski kaganat, Belgrade 1977, p. 12. See also J. Vásáry, «Runiform signs on objects of the Avar period (6th-9th cc. A.D.)», Acta Orientalia 25, Budapest 1972, 335-347.


(6) Most famous and also historically important are the T'u-chüe (Türküt) Turkic imperial runic inscriptions from the Orkhon valley (Mongolia; first half of the eighth century). They are given with English translations (not always reliable) by Talât Tekin in his A Grammar of Orkhon Turkic, Bloomington, Ind. 1968, pp. 231-295.





ten records about their goals or achievements. The second reason is that scholars studying the Avars have not exploited fully the multilingual and heterogeneous sources for the history of the Eurasian steppe, since it is very difficult to combine experience in both western and eastern philologies in order to put together, for the first time, a historical perspective of the nomadic steppe empires [7].


In this paper I will try only to take a fresh look, through the perspective of Eurasian and general history, at the Avars in their relationship to the Slavs. Let us first deal, if only briefly, with the vexed question as to whether the European Avars were identical with the East Asian Jou-jan of Chinese sources [8]. Chinese official historiography of the ancient and early medieval period used two generic designations for «barbarians» to the northwest: Hsiung-nu and Tung Hu. The Tung Hu or «Eastern barbarians» were known from the third century B.C.E., and later developed two branches: the Wu-huan, first mentioned in 78 B.C.E., and the Hsien-pi, documented from 45 C.E. Chinese historical phonology, which is now a precise and reliable discipline [9], allows us to reconstruct the ancient



(7) Three more or less reliable attempts to place the Avars in universal history have been published recently, but they are limited by a Europe-centered attitude. One, by a Slovak historian with some Byzantinological background, Alexander Avenarius, was cited in note 2, above. The second, by a leading Serbian archeologist, Jovan Kovačević, Avarski kaganat (Belgrade 1977), gives an evenhanded presentation of all known basic archeological facts, but since it lacks footnotes, a non-specialist reader may have difficulty in evaluating it. The best of the three, however, is the short but well documented overview by Wojciech Szymański, cited in note 2, above.


(8) See William Samolin, «Some notes on the Avar problem», Central Asiatic Journal, 3, The Hague 1957-58, 62-65, esp. 62.


(9) The basic works are: Bernhard Karlgren, Analytic Dictionary of Chinese and Sino-Japanese, Paris 1923; Id., Grammata Serica Recensa, Stockholm 1957; W. A. C. H. Dobson, Late Archaic Chinese, Toronto 1959; Id., Early Archaic Chinese, Toronto 1962; Edwin G. Pulleyblank, «The consonantal system of Old Chinese», Asia Major 9 : 1, London 1961, 58-144; 9 : 2 (1963) 106-165; Id., «An interpretation of the vowel systems of Old Chinese and Written Burmese», Asia Major 10 : 2 (1963) 200-221; Id., «Late Middle Chinese», Asia Major 15 (1970) 197-239; 16 (1971) 121-168; Sergej Evgenievič Jaxontov, Drevnekitajskij jazyk, Moscow 1965.





pronunciation of the two designations: these are *ahwar (= Avar) for the Wu-huan, and *säbir, säbär (> Sibir, hence Siberia) for the Hsien-pi [10].


The leading clans of both the Jou-jan in the steppe and the Tabgach (T'o-pa) Wei dynasty in China (386-057) originated from the Hsien-pi and both used a Proto-Mongolian language as their lingua franca [11]. The European Avars were not directly connected with the real Avars, but, as contemporary Byzantine sources clearly state, consciously imitated them, especially by copying the way they plaited their hair [12], in order to gain for themselves the prestige the true Avars, the Wu-huan of Chinese sources, had enjoyed among the steppe peoples. It is also clear, especially from the Byzantine data, that these Pseudo-Avars were of Hunnic origin, the Vär-Hunni (Οὐάρ Χουννί) [13]; I will return to this topic below. It is this group, which will henceforth be called simply Avars in this paper, that is important for the history of Europe. They have nothing to do with the Asian Jou-jan [14].




(10) For details, see O. Pritsak, Studies in Medieval Eurasian History, London 1981, study no. VI, pp. 157-163. E. G. Pulleyblank was the first to show that Wu-huan is the equivalent of Avar, in Asia Major 9 : 1 (1963) 259.


(11) See Louis Ligeti, «Le Tabgatch un dialecte de la langue Sien-pi», Mongolian Studies, Budapest 1970, 265-308. Cf. Peter Olbricht, «Uchida's Prolegomena zu einer Geschichte der Jou-jan», Ural-Altaische Jahrbücher 26, Wiesbaden 1954, 90-100.


(12) Source data are given by A. Kollautz and Hisayuki Miyakawa, Geschichte und Kultur eines völkerwanderungszeitlichen Nomadenvolkes: Die Jou-jan der Mongolei und die Awaren in Mitteleuropa, vol. 2, Klagenfurt 1970, pp. 114-121.


(13) See the clear statement of Theophylact Simocattes (fl. 610-641), ed. Hans Wilhelm Haussig, Byzantion 23, Bruxelles 1954, 283-284.


(14) The two volumes by Kollautz and Miyakawa (cf. note 12 above) failed to prove that the two groups were identical. Cf. also Hans Wilhelm Haussio, «Zur Lösung der Awarenfrage», Byzantinoslavica 34, Prague 1973, 173-192.








There were two kinds of «nomads»: those of the steppe and those of the sea.


The Avar realm in Europe was one in a sequence of Eurasian steppe «nomadic» empires, which had a series of fundamental traits in common [15]. Let us examine the general structure of such an empire, for which I use the term pax, reserving empire to refer to sedentary societies. The pax was the creation of the bearers of a steppe warrior religion («Männerbünde», «Royal Hordes») characterized by specific notions to which we will return.


The idea of rule by professional warrior and merchant elites emerged on the territory of present-day Mongolia about 900 B.C.E., probably among Indoeuropeans (Iranians, Tocharians). Some five hundred years later the idea was taken over by the Altaic-speaking peoples, who made use of it for almost two millenia. The first to do so were the Hsiung-nu, who competed with the Tung Hu or Eastern Barbarians, especially the Hsien-pi or Säbir subdivision.


A special characteristic of the pax was a lack of linear thinking, of the concept of linear development. Time and events were viewed as cyclic. A particular clan could acquire charisma, rise to power, and rule for a certain period, only to be swept away by another group. Whenever a charismatic clan lost charisma, not only its ruling position, but also its name and language were replaced



(15) On the contrast between sedentary and nomadic empires, see O. Pritsak, The Origin of Rus', vol. 1, Cambridge, Mass. 1981, pp. 10-20. The Tängri (Täŋri) religion is discussed there on pp. 14, 18, 73-82. The concept of the inner and outer territories is elaborated in O. Pritsak, «Where was Constantine's Inner Rus'?», Okeanos = Harvard Ukrainian Studies (abbrev. HUS) 7 (1983).





by those of the victorious clan. This does not mean that all the people immediately shifted from one language to another, only that their allegiance changed, and with it the relative prestige of different languages. As soon as one cycle was completed, another followed, in a round of events preconditioned by the society's non-linear, cyclical thinking.


The territory of the pax was also exchangeable; what was indispensable was that the rulers have people who would follow their lead. For the territory itself there were only two requirements: first, that it was located along an important commercial route, and second, that it was close enough to a sedentary empire to tap its economy by periodical raids, yet distant enough to allow the warriors to disappear into marshes, sands, or forests. Thus, for example, warriors from Mongolia could easily cross the familiar deserts to raid China more or less at will, but the Chinese found it almost impossible to trace the enemy through the wilderness to the inner stronghold. The sea nomads based, e.g., on the islands of Denmark, could thread their way through dangerous passages to mount a raid against any coastal settlement, but the people they attacked would have the greatest difficulty in following them back to their base.


In consequence, the territory of the pax was divided into inner and outer areas, each with its own admi-nistration, troops, and offices, so that every function existed in pairs,  inner and outer.


The inner area, such as the sandy desert of the Orkhon Valley in Mongolia, was the place where the religious and imperial sanctuaries were found. From the time of the T'u-chüe Pax, 550-744, on, the sanctuaries were also decorated with dynastic inscriptions. Here the annual rites and ceremonies of the Tängri religion were held. The





inner area, moreover, served as the training ground for new military recruits. The outer area for any pax was the frontier (limes) of the empire of the time (China, Iran, Rome), where the nomads could station garrison-type military units, specially trained for local conditions.


This area was of crucial importance for the pax, since it was where contacts were made with the civilization, culture and, above all, the economy of the sedentary empire. This contact-area is also a focal point for the historian, since it was only there that the non-historical steppe polities entered the stream of history.


At the time and place of concern to us - sixth century Europe - the outer area was the Roman (Byzantine) Danube Limes and the Frankish Saale-Elbe frontier: therefore it is events on these territories that will occupy our attention.


The charismatic clans themselves were not pastoralists, but professional warriors, ideologists and leaders [16]. Before coming to power, their members usually lived in towns [17], where they gathered knowledge about economic, geographic, and political affairs and conditions. They maintained close ties with the international merchants, and often served them as guards in their emporia. The (itinerant) merchants, on their part, were interested in the ongoing existence of a pax which could insure peaceful



(16) By «professional», I mean that they had training and probably some experience, along with charisma, and therefore were known as people with special expertise. To take an analogy from recent history, the newly-independent Balkan nations of the nineteenth century, after experimenting with local rulers, looked to known «charismatic clans» – the families of Saxe-Coburg, Oldenburg, Hohenzollern, and the like — to come, with retinues of professional administrators, to organize the new states of Bulgaria, Greece, Rumania, and Albania.


(17) The Chinese sources uniformly connect the A-shi-na, the charismatic clan of the T'u-chüe/Turks with the city-oasis of Turfan (Kao-ch'ang), see Chou-shu, chap. 50, fol. 1, and Sui-shu, chap. 84, fol. 1.





conduct of their business affairs. During the intermezzo between two cycles of ruler-clans, the merchants would search for a suitable new ruler. When they found one, they would supply him with money, arms, and provisions. Once the pax was established, the merchants would function within it as an organized body of tax-farmers and duty-free traders.


The ambitious leader of a suitable clan who wished to become the ruler of a pax first had to gather together a retinue of the Germanic  comitatus  type, that is of professional warriors who were personally selected by the leader and therefore owed him special allegiance. The new pretender, having established such a retinue (in Old Turkic buyruq, in Hunnic boyla-r) , would set out to win decisive victories, in order to attract attention. If he was successful, adventurers from far and wide would flock to him in expectation of sharing in an El Dorado. The territorial center of nomadic society, e.g. Mongolia (n. of the steppe), or the islands of Denmark (n. of the sea) would be flooded with newcomers. There, in what Jordanes called officina gentium [18], the recruits would undergo a period of training before riding out in successive waves that often resulted in a migration of peoples.


The new leader's next step would be to find cohorts among the leaders of other charismatic clans or important pastoralist tribes. A system of intermarriages would develop of necessity, because in the steppe exogamy prevailed. With the help of his new partners, the steppe ruler would take control, either by force or by persuasion, of the human resources needed to assure a steady supply of military recruits and revenues for the upkeep of the army and administration. For a nascent steppe pax, the



(18) See sect. II. 8 of this paper.





ideal sources were the pastoralist tribes and then the hunting tribes, because of their mobility and expertise with weapons. However, if the territories were unsuited for pastoralism, especially in the outer area (limes), other arrangements could be substituted. Chief among them were the system of military settlements (of the type of the later Russian Kazak settlements) and the military slavery system like that practiced in Egypt and Syria during the Mamlūk era, or the Ottoman qulluq system, or some combination of these arrangements appropriate to the situation [19].


Foreign administrative expertise was also required. Usually suitable experts were found among defectors or individuals kidnapped from the sedentary societies.


The most powerful tactical military unit employed in the steppe was a tümän, or a 10,000 man unit, usually of cavalry. Each househould, that is five to seven people, was expected to provide two or three adult males, over twelve years old. The territory of the pax was divided into military-political districts called «arrows» (Old Turkic oq, pl. oġəz, later -ġuz, Hunnic pl. oġur, later -ġur) based on the capacity of its population to supply the ten thousand men necessary for a tümän. In practice, this required that a large ethnic tribe be subdivided, or that several smaller ethnic units or tribes be joined into one military «arrow». For the oq-system to function, the apolitical anthropological tribe (Old Turkic oqsəz) had to acquire appropriate military training and political indoctrination. Camaraderie and uniform military training brought about linguistic assimilation; first, dialectal differences among men of a tümän would fade, and then



(19) See the recent study by Daniel Pipes, Slave Soldiers and Islam, New Haven and London 1981.





the emerging lingua franca of the pax would take the place of other languages still used among the specific tribes. The prestige of the warrior, whether on duty or in retirement, prompted the rest of the population to imitate his speech. The Turkic pax was established in Mongolia in 550; by 576 it already reached as far west as Tamatarcha (later T''mutorokan') at the Kerch Straits. However, it was some time before a supratribal Turkic medium of communication could develop. At first, the official chancery used East Iranian Sogdian, as witnessed by the newly discovered Bugut inscription from about 580 [20]. We know Old Turkic written texts from no earlier than about 700, give or take a couple of decades [21]. Their language, the oldest version of Turkic, was already without important dialectal differences; there is no trace of dialectal peculiarities that must have existed before this period of a Turkic lingua franca. Not until after the dissolution of the Turkophone pax in 845 did separate Turkic languages and dialects begin to emerge. This is a situation comparable to the emergence of the Romance languages after the fall of the Roman Empire, which had used Latin as both its lingua franca and its chief medium for cultural expression [22].


During the period between the fifth and tenth centuries,



(20) Sergej G. Kljaštornyj and Vladimir A. Livšic, «Sogdian inscription of Bugut revisited», Acta Orientalia 26, Budapest 1972, 69-102.


(21) See Louis Bazin, «La litterature épigraphique turque ancienne», Philologiae Turcicae Fundamenta, vol. 1 Wiesbaden 1964, pp. 192-211.


(22) For Latin, we know that a few traits later characteristic of individual languages existed fairly early, and we must assume this to be true also of Turkic; what is important is that the major structure and the most important details appear to be shared in all areas. As for Greek, the initial establishment of the koiné, the lingua franca of the Hellenistic age and the ancestor of modern Greek, was accomplished during the hundred years after the death of Alexander of Macedon. For a good account, see Karl Brugmann and Eduard Schwyzer, Griechische Grammatik, Munich 1953, pp. 116 ff., and Albert Debrunner, Geschichte der griechischen Sprache, vol. 2, Berlin 1954.





the Eurasian steppe was the home of two main pastoralist groups, probably of heterogeneous origin. When they were organized into the frameworks of paxes, two linguae francae developed, one Turkic and used in the east, the other Hunnic and used by the western group. This is reflected in the recorded names of some of the groups. In the names of the political tribal groups - usually organized in two wings, left and right - the numerals 5, 9, 10 and 30 occur. The corresponding Hunnic words were *bīl (later bŭl) ' 5 ' (Turkic bēš), *qutur ' 9 ' (Tu. toqəz), *onno ' 10' (Tu. ōn) and *utur ' 30' (Tu. otəz). Therefore the Old Turkic runic inscriptions from Mongolia mention tribal groups such as the Toqəz Oġəz and On Oq. [23] and characterize them as either loyal members of the Türküt-led confederation or as rebels striving for independence, sometimes (the greatest sin) in cooperation with China. At the same time, the Western, mainly Byzantine, sources often mention the Onnoġur, Quturġur, and Uturġur [24] as unruly members of both the (Proto-)Bulgar and the Avar confederations. I mentioned earlier that the paxes usually employed special garrison troops along their frontiers, that is in the outer zone. They too were structured in groups, whereby the same numbers were used in their names as in the «arrows». While the Chinese sources contain such designations as Chiu-hsing Hu, «The Nine Tribes of the Iranian [military settlers] [25]», the Byzantine sources mention «the Seven Tribes» of the so-called Sklavins [26] within the pax led by the Proto-Bulgars.



(23) T. Tekin, Grammar, see fn. 6 above, pp. 232 (IS12) and 233 (IE14).


(24) The Byzantine data on these groups are conveniently listed by Gyula Moravcsik in Byzantinoturcica, 2nd ed., 2 vols., Berlin 1958, sub voce.


(25) Cf. E. G. Pulleyblank, «A Sogdian colony in Inner Mongolia», T'oung Pao 41, Leiden 1952, 317-356.


(26) τὰς λεγομένας ἑππὰ γενεάς, Theophanes (d. 818), «Chronographia», ed. Igor' Sergeevič Čičurov, Vizantijshie istoričeskie sočinenija: «Xronografija» Feofana, «Brevarij» Nikifora, Moscow 1980, p. 37. On the term γενεά '1. gens; 2. generatio ', see Čičurov's commentary, pp. 120-121 (fn. 300). See also Wincenty Swoboda, «Siedem plemion», Słownik starożytności słowiańskich, vol. 5, (Warsaw 1975), pp. 157-158, with bibliography, and the study by Ivan Dujčev cited in fn. 166.





Therefore we must take into account that 'tribe' may, in these sources, in fact mean oq, thus a military rather than an ethnic designation.





Shortly after 450, the Chinese Wei dynasty (the T'o-pa, Tabgach, of Proto-Mongolian origin) put an end to the state of Pei-Liang, the last of the Hunnic commercial centers created by the Hsiung-nu on Chinese territory, in the economically vital province of Kan-su. Some of the defeated ruling clans managed to flee to the distant Hsiung-nu successor state near Lake Balkash, an area known as Yüe-pan in the Chinese sources [27], reflecting *r-pän as the old pronunciation [28]. This same designation also occurs in the Old Turkic Bilgä Qagan inscription of 732 (II E 20) [29].


A century later, the Turks (T'u-chüe) proclaimed their pax in Eurasia (550), and some of the young people of the rpän-Hunnic group, to the number of about twenty thousand (two tümäns), fled to the European frontiers of the Byzantine empire. Theophylact Simocattes (fl. 610-641) called them, properly, Οὐάρ Ouar (= *r > vär) and Χουννί Chounni, but he also wrote that these refugees pretended to be the true Avars, because of the latter's



(27) See the monograph on Yüe-pan in Pei-shi, ch. 97, fol. 15-16.


(28) According to Karlgren's Analytic Dictionary (henceforth AD, cf. fn. 9) the old pronunciation of the signs nos. 1138-690 was wät-puân, i.e. *rpän.


(29) I cite the Orkhon inscriptions (I = Kül Tigin, A.D. 731; II = Bilgä Qagan, A.D. 732) according to the «Finnish atlas»: Inscriptions de l’Orkhon, Helsingfors 1892.





authority among the steppe peoples [30]. Among the first to submit to the Pseudo-Avars were the Iranian merchant clan of Warāz/Barč (Βαρςήλτ) [31], the Hunnic Onnoġurs (Οὐννουγοὺροι), and the Proto-Mongolian Säbirs (Hsien-pi = Σάβιροι) [32]. The victorious refugees established themselves in the Northern Caucasus, near the Byzantine holdings in the Crimea. In 558, through the good services of the Alan ruler, they established relations with the Byzantines and soon were granted the status of foederati on Byzantine territory in Scythia Minor, that is, Dobrudja [33].


We have now arrived at the non-controversial stage in Avar (Pseudo-Avar) history, for the newcomers from Asia, having established their «outer territory» on the Roman Danube limes, had become involved in their colorful relationship with the historically-minded Byzantines and their activities are fairly clearly recorded. Let us therefore turn to the Slavic side of the issue.



(30) See Theophylact Simocattes, as quoted in fn. 13. Menander Protector (scr. 583-585) uses Οὐαρχονῖται, to refer to them (ed. L. Dindorf, HGL, vol. 2 [Leipzig, 1871], pp. 86-87), a form containing the suffix /it/. - On the Hunnic form *vär < *r see O. Pritsak, «Ein hunnisches Wort», Zeitschrift der Deutchen Morgenländischen Gesellschaft, 104, Wiesbaden 1954, 124-135.


(31) On this clan, see O. Pritsak, «The Khazar kingdom's conversion to Judaism», HUS 2 (1978) 261.


(32) See Theophylact Simocattes, as in fn. 13. For Chinese Hsien-pi as equivalent to Byz. Σαβιρ-/Σαβηρ- see O. Pritsak, «From the Säbirs to the Hungarians», Hungaro-Turcica. Studies in honour of Julius Németh, Budapest 1976, pp. 28-30.


(33) See Menander Protector, «Excerpta» (ca. 584), ed. Ludwig Dindorf, Historici Graeci minores, vol. 2, Leipzig 1871, p. 4.




[Back to Index]