Studies in the History and Language of the Sarmatians
J. Harmatta


1. A History of the Problem

After Müllenhoff's fruitful activity [1] it was Miller's investigations [2] that produced a great advance in the research on the language of the Iranian tribes in South Russia. The ancient inscriptions of the Pontic region were collected and edited by Latyshev, [3] so that the fairly large number of names appearing in the inscriptions has become easily accessible to linguists. Miller had made a thorough study of Ossetian, a language still spoken in the Caucasus, and on the basis of his expert knowledge of that tongue, he began to investigate the material of names appearing in the ancient inscriptions of South Russia. His work was crowned with success: with the help of Ossetian, he managed to find out the meaning of a considerable portion of the non-Greek names in the inscriptions. The phonemic form of the names thus interpreted by Miller shows, in many cases, a phonemic development parallel with that of Ossetian. These correspondences may be summed up as follows:

1. The initial phoneme p- of the Old Iranian languages has a corresponding  f both in the names figuring in the inscriptions and in Ossetian: e. g. = Ossetian fidä 'Vater' ~ Avestan pitā = Ossetian furt 'Sohn' ~ Avestan puϑra-, etc.

2. The Old Iranian initial group of phonemes fri- developed into li-:  = Ossetian limän 'Freund' ~ Old Iranian *friyamanah-.

3. Old Iranian initial v- has disappeared before i = Old Ossetian *insadz-ag, cp. Western Ossetian insäi, Eastern Ossetian ssäj 'zwanzig' ~ Avestan vīsaiti.

4. Old Iranian initial h- has disappeared before a (in the following word: ) = Ossetian avd 'sieben' ~ Avestan hapta-.

5. Instead of Old Iranian r we find l before i: see above  and also , in which  = Ossetian fäl- ~ Avestan pairi.

6. The Old Iranian initial group of phonemes ary- developed into ir-:  = Ossetian ir 'Ossete', iron 'ossetisch' ~ Avestan airya-.

7. In place of the Old Iranian group of phonemes ti we find the groups ts or dz = Old Ossetian *insadz ~ Avestan vīsai-ti.

8. The Old Iranian group of phonemes -ϑr- is replaced by –rϑ- or -rt-: 

Additional Notes

To p. 58. In connection with my views regarding the language of the Sarmatians an argument arose in two essential points: 1. the family-tree theory, 2. the problem of the Sarmatian dialects and their relation with Scythian on the one hand and Alanic and Ossetian on the other hand.

As to the first point I should like to emphasize that rejecting the family-tree theory I did not call in question the genealogical relationship of languages. But I insist on the statement that the family-tree theory presents no suitable model for the real development of languages. This was, perhaps, overemphasized by me at that time but in view of the remarks made by V. I. Abaev and I. Gershevitch on this subject, I think even now that it was necessary.

My esteemed friend Professor V. I. Abaev regards all Northern Iranian linguistic materials (including Scythian, Sarmatian, Alan and even Saka names) as Scythian (. I. Moscow—Leningrad 1949. 148). Moreover, he thinks that it is "to force an open door" if one tries to prove the existence of Sarmatian dialects on the one hand, but at the same time assuming that all data only reflect different stages of the same phonemic development leading uniformly to Ossetian, he actually denies the existence of different Sarmatian dialects on the other hand. The same opinion was expressed by W. P. Schmidt, BzN 7 (1956) 209 foll., while L. Zgusta wanted to reduce the number of the dialects reflected by the Iranian names occurring in the Pontic Greek inscriptions to two and to regard these as two different languages, viz. Scythian and Sarmatian, instead of various dialects of the Sarmatian (Die Personennamen griechischer Städte der nördlichen Schwarzmeerküste. 245 foll., Acta Orient. Hung. 4 [1954] 245 foll.). It seems that sometimes it is not easy to force even an open door.

Hoping that I shall still have the possibility to publish a detailed analysis of all Scythian, Sarmatian and Alanic linguistic materials, I would only insist on the following points.

1. From historical view-point it is impossible to regard all Iranian tribes and languages of Eastern Europe and even Central Asia as Scythian. This would mean a return to the linguistic usage of the ancient and mediaeval historians and geographers who denoted practically all Eastern peoples, even the Hungarians as Scythians. As I pointed out almost three decades ago (Quellenstudien zu den Skythika des Herodot. Budapest 1942.), among the ancient authors it is only Herodotos who clearly differentiated the Scythians from the other Northern and Eastern peoples on the basis of the language and customs. It is, therefore, only correct to regard the tribes denoted by Herodotos as Scythian ones and their historical descendants as Scythians.

2. As to the "open door" I would only remark that my humble paper was the first attempt to prove the existence of different dialects (or languages) within "Sarmatian". It makes a difference to guess something or to prove it.

3. The argumentation that the linguistic differences reflected by the Iranian names occurring in the Pontic Greek inscriptions only represent different stages of the same linguistic development leading uniformly to Ossetian, is a simple sophistry. With similar argumentation one could assume that Parthian developed into Persian because in comparison to the latter it preserved an earlier stage of linguistic development in most cases. The only fact we can establish is the existence of linguistic differences reflecting various dialects (or languages), while the assertion that all Sarmatian languages (or dialects) developed uniformly into Ossetian, is a mere invention without any real basis.

4. The theory expounded by Zgusta cannot be accepted for the following reasons:

a) Neither did he study the geographical distribution of the linguistic data thoroughly enough nor did he qualify the names according to the various Scythian and Sarmatian peoples. One must not operate with such vague concepts as "Scythian" or "Sarmatian" at that late epoch.

b) In the 2nd and 3rd centuries A. D., i. e. at the time to which the bulk of the names is to be dated, we can no more speak of Scythians.

c) Zgusta did not take into consideration either the interrelation of the phonemic changes jointly occurring in the same names (cf. my Studies in the Languages of the Iranian Tribes in South Russia. 55 and here p. 96 above), or the fact that evidences for different dialects frequently occur in the one and same Greek city.

d) He tried to eliminate all linguistic data, e. g. even the name Alani, which contradict his theory of the existence of only one Scythian and one Sarmatian languages in the North Pontic region in the 2nd-4th centuries A. D. This is, of course, inadmissible. For further details I refer to my forthcoming study mentioned above.

5. Gershevitch thinks that the dialect differentiation of the Sarmatian "cannot be assigned to particular areas or tribes" and some of the differencies "may not be synchronic, but due to diachronic sound-change within one dialect". Contrary to his assertion Zgusta has proved in any case that at least a great part of the dialect differencies can be assigned to particular areas. As to the "diachronic sound-change" I refer to my remarks above.

6. Gershevitch does not understand against what I was arguing when discussing the character of the Sarmatian language, because in his opinion nobody to-day takes the term "family-tree" literally and to speak of the Sarmatian or Alanic language instead of group of languages, "is merely a convenient simplification". I do not want to discuss here whether any simplification — convenient or inconvenient — of the reality can be regarded as correct or to insist on the question what means to take the family-tree theory (not term!) literally or not literally. I only refer to the fact that Gershevitch admits the dialectal differentiation of Proto-Iranian theoretically, at the same time, however, he adheres to the purely formal linguistic concept of Proto-Iranian as one can state on the basis of his remarks regarding Proto-Iranian *puθra-. Thus he does not realize that these two conceptions are irreconcilable with each other. Proto-Iranian as a formal linguistic concept means the total of the common linguistic features which can be deduced by the comparison either of the Old (sometimes Middle or even Modern) Iranian languages or of Old Iranian and Old Indian. In this sense Proto-Iranian must, of course, represent a linguistic unit because precisely the unity of the common features is the logical postulate of its existence. This Proto-Iranian is, however, only a linguistic abstraction which can never correspond to reality. Therefore. I proposed to reinterpret Proto-Iranian and other similar purely linguistic concepts from the historical point of view and to replace these abstractions without space and time by the reconstructions of historically definable languages or linguistic states. This was discussed by me Studies in the Language of the Iranian Tribes in South Russia 22 foll. at lenght. Apparently Gershevitch did not understand my argumentation and he confused the linguistic and historical concepts of Proto-Iranian. That was exactly against what I was arguing. It is interesting to observe that after all Gershevitch, BSOAS 17 (1955) 486 joined to the theory according to which the Alans "were brought from the area of Lake Aral to the Caucasus by a migratory movement". This conception implies, of course, that the Alans have nothing to do with the Scythians and the Sarmatians.

1. Deutsche Altertumskunde, III. 101—125.

2. His chief works: , I-III.  1881-7;  1902; Die Sprache der Osseten, Strassburg 1903: Ossetica 1904; the Ossetian dictionary published posthumously by Freimann: , I-III.  1927—34.

3. Inscriptiones antiquae orae septentrionalis Ponti Euxini graecae et latinae, I. II., IV. 1885—1901.


= Ossetian furt 'Sohn' ~ Avestan puϑra-;  in names like  etc. = Ossetian äχsart 'Macht' ~ Avestan -χšaϑra-.

9. The Old Iranian group of phonemes -χr- developed into –rχ-:  = Ossetian surχ 'rot' ~ Avestan suχra-.

10. The numerous names ending in  appearing in the inscriptions, correspond exactly to the present active participles formed in Ossetian with the ending -äg: e. g. = Ossetian iγosag 'gut hörend, guter Hörer', from the verb γos-un 'hören';  = Ossetian käsag 'guter Seher' from the verb käs-un 'sehen, schauen'. (In this case we come up against an obvious error of Miller's, since the words iγosag and käsag contain, not the ending -äg mentioned alove, but the suffix -ag, -agä which is used to form adjectives expressing permanent qualities from the present stem. The forms of names appearing in the inscriptions may, of course, just as well stand for adjectives formed with the suffix -ag as for participles formed with the ending -äg).

11. The suffix , found in some of the names in the inscriptions, corresponds exactly to the suffix -gin which forms adjectives in Ossetian:  = Ossetian nom-gin 'namhaft, berühmt'. [4]

On the basis of these correspondences Miller came to the following conclusions:

1. The Ossetes belong to the Iranian group of the Indo-European family of languages.

2. The ancestor of the Ossetian language was one of those dialects which had developed in the northern part of the territory once inhabited by the Iranians, i. e. on the steppes of Central Asia, lying roughly to the north of the rivers Oxus and Yaxartes.

3. The separation of this dialect from the common Iranian parent language had taken place in prehistoric times, before the cultured nations of Iran — the Medes and Persians — entered the course of their historical existence.

4. The ancestors of the Ossetes belonged to those nomadic Iranian peoples who, for many centuries, were known partly as Sarmatians and partly as Scythians, and who occupied the steppes stretching along the Pontus and the Sea of Azov. [5]

From this formulation of Miller's it does not appear clearly whether, in his opinion, the dialect, which he regards as the ancestor of the Ossetian language, was the common dialect of the Scythian—Sarmatian tribes, or a separate Ossetian dialect quite apart from the tongue of the Scythian—Sarmatian tribes. From Miller's other remarks, about the position of the Ossetian language, it appears, however, that on the whole he regards the Scythians and Sarmatians as the ancestors of the Ossetes and that, in his view, the language of the Pontic Iranians (Scythians and Sarmatians) must be identified with Old Ossetian, i. e. an earlier stage in the development of the Ossetian language. [6]

After Miller, it was Vasmer who dealt in some detail with the language of the Iranian tribes in South Russia, [7] in a much more cautious manner. This caution is especially noticeable when he discusses the mutual relationship of the available Scythian and Sarmatian names. Vasmer has attempted to separate, on the basis of the available material of names, the language of the Scythians, from that of the

4. See , III. 83, Die Sprache der Osseten, 6 foll. With regard to par. 6 see Vasmer, Die Iranier in Südrußland, 28.

5., III, 100 foll, and also 73.

6. See e. g. , III, 101: ; Die Sprache der Osseten, 7: „Diese Eigentümlichkeiten der politischen iranischen Sprache gestatten uns, in derselben eine Vorstufe der Ossetischen zu sehen, welche als ein Nachkomme der ausgestorbenen 'Sarmatisclien' gellen kann“. See also ibid. 4, 5.

7. Untersuchungen über die ältesten Wohnsitze der Slaven. I. Die Iranier in Südrußland, Leipzig 1923, Iranisches aus Südrußland: Streitberg-Festgabe: 367—376, and also RLV XII, 236—251.


Samatians. But he has no doubts, either, as to the close connection existing between Sarmatian-Alanic, on the one hand, and the Ossetian language, on the other. [8] His formula admits of a wide range of possibilities.

Vasmer's caution was undoubtedly well-founded since, though it is possible that ethnically the Ossetes are the descendants of an Iranian tribe in South Russia, it is hardly likely that a strikingly large number of Iranian tribes from South Russia, appearing in different places and under different names in the course of history, could be gathered into a single unit. Neither is it likely that their language could be regarded as Old Ossetian, i. e. as an earlier stage of the present Ossetian language. Vasmer's attempt to separate the language of the Scythians from that of the Sarmatians was not very favourably received. The negative attitude to Vasmer's theory found its clearest expression in Lommel's criticism. The latter admits the possibility of linguistic differences between Scythians and Sarmatians but, according to his view, these must have been quite insignificant. Against the differences which, in his opinion, cannot even be demonstrated, Lommel emphasizes those linguistic peculiarities of Scytho-Sarmatian which closely connect this latter group of languages with Ossetian and Sogdian. Such is the use of the -t as the plural suffix in all these languages (Scythian—Sarmatian—Alanic , Ossetian -, -t’ä, Sogdian -t). In Lommel's view this way of forming the plural may date from very early times, and may have spread very long ago over the whole linguistic area of Northern Iran. Thus in Lommers conception the picture of different Northern Iranian languages or dialects is replaced by a homogeneous Northern Iranian linguistic community or linguistic area. [9]

The idea of a Northern Iranian linguistic group that forms the background of Lommel's arguments took definite shape only after the important archaeological discoveries in Eastern Turkestan had brought the Sogdian language to light. It was at this time that, following Andreas' hints, Gauthiot formulated his theory, according to which Sogdian, Chorasmian, Alanic, and Ossetian, together with the rest of the related languages, formed a common "Scythian" group of languages. [10] Gauthiot's theory found, on the whole, general acceptance. One of the most prominent common features of this "Scythian" group of languages is the formation of the plural with -t, already referred to above [11]; after Tomaschek [12], Marquart [13], Lommel, Jacobsohn [14] and some other scholars it was Kretschmer who recently tried to prove the existence of this feature, on the basis of a more detailed argumentation from the Scythian language, with the plea that its presence in Yagnobi, Sogdian, and Ossetian argued for its extreme antiquity. [15]

The basis of all these conjectures and arguments is formed, whether consciously or anconsciously, by the old theory of the family-tree of languages. According to this theory, the Aryan branch, having become independent of the primitive Indo-European linguistic community, was only gradually divided into Indian and Iranian, Iranian in

8. Die Iränier in Südrußland, 28 foll.

9. See AfslPh XL (1926), 151 foll.

10. Essai de grammaire sogdienne. Vol. I, Paris 1914—1923, III.

11. See Benveniste, Essai de grammaire sogdienne, Vol. II. Paris 1929. 79.

12. SWAW CXVII (1888), 47.

13. Untersuchungen zur Geschichte von Eran, II, Leipzig 1905, 78 foll.

14. KZ LIV (1962), 268.

15. Glotta XXIV (1936), 42. — The ending  in Scyihian-Sarmatian tribal names was first compared by Miller with the Ossetian plural sign -. Vasmer was the only scholar to reject this explanation (Iranisches aus Südrußland: Streitberg-Festgabe, Leipzig 1924, 373 foll.) but his arguments were found unconvincing by all sholars, including H. W. Bailey (Asica, reprinted from TrPSh (1945), 25 foll.). Nevertheless, the question requires fresh, more detailed examination.


its turn being subdivided later into the Northern (or "Scythian"), Southern, Western, etc. branches. Anyone imbued with the spirit of this theory would naturally attribute the common features in different languages to an ancient unitary linguistic community; the farther he travels back on the road leading from individual languages, to the original linguistic community, the less inclined he becomes to assume the existence of linguistic or dialectal differences in the languages of human communities. This explains why Miller tried to establish the following line of development: Scythian — Sarmatian — Alanic — Ossetian, why Lommel thought it unlikely that there were any tangible dialectal differences within the North Iranian or "Scythian" branch, why the plural formation with -t was attributed to such an early date. Seen from the angle of the family-tree theory, the linguistic facts could be best explained by assuming the former existence of a "Scythian" branch speaking a uniform language, and developing, through a slow process of differentiation, into languages like Ossetian and Yagnobi, still spoken to-day. Starting from the premises of such a theory one naturally could not assume the existence of any noticeable dialectal differences in the various groups of Scythian and Sarmatian, since these languages represented an earlier stage in linguistic development.

The limitations imposed by the family-tree theory upon research may be best observed in Vasmer's case. He already noticed that in the material of names figuring in the inscriptions there are forms bearing witness to different lines of phonemic development. In some instances, when the forms were obviously synchronous and differences could not be explained as being due to temporal succession, he actually thought of these differences concealing some dialectal variety. In most cases, however, he did not reach this conclusion, but either disregarded facts testifying to the existence of dialectal differences, or tried to assign such forms to a later date [16].

A similar theory also underlies Sköld's researches into the Ossetian loanwords in Hungarian, and the related problem of Ossetian dialects. Sköld tried to prove that the Ossetian loan-words in Hungarian derive, not from an extinct Alanic or Ossetian dialect, but from Eastern Ossetian which is still a living language. In his view the Ossetes and the Alans formed a single people who once used to inhabit a large territory. Nevertheless, he thought it impossible to assume the existence of other Ossetian dialects at an early date, apart from those two which are still spoken. Thus in Sköld's theory, too, we are clearly faced with the idea that we cannot assume a greater linguistic differentiation than that prevailing at the moment [17].

Sköld's conception is based on the mechanical and forced application of a theory: it is best shown by his disregarding the fact that even present-day Ossetian has more than two dialects. Already Miller noticed three Ossetian dialects (Western, Eastern, and Southern Ossetian) [18]. Recently Abaev's investigations have clearly demonstrated that in the Southern Ossetian territory alone there are three separate dialects, easily distinguishable by their phonemic characteristics. [19] If Sköld had no doubts with regard to the existence of the eastern and western Ossetian dialects as early as the age of linguistic connections between Ossetes and Hungarians, he naturally could have no reason to doubt the existence of other Ossetian dialects in the same period. So he simply paid no attention to the southern Ossetian dialect or dialects which contradicted his theory.

16. Iranisches aus Südrußland, 370.

17. ZII III (1925), 179 foll., Die ossetischen Lehnwörter im Undarischen. Lund — Leipzig 1925, 66 foll.

18. Die Sprache der Osseten, 2.

19.  87 foll.


Thus it is entirely natural that Sköld's conclusions about the Ossetian loan-words in Hungarian, and the relations between Alans and Ossetes in general, have been recently thoroughly revised by Abaev. Abaev refuses to view the problem of Alanic — Ossetian contacts as a problem of racial and anthropological relations, he regards the Alans simply as "forebears", the Ossetes as "descendants", as Miller had done. In his view, the question of Alans and Ossetes is significant only as the "problem of cultural-historical and linguistic contacts between two peoples of the Northern Caucasus, one of them living at the present time, the other in the Middle Ages." [20]

Abaev has sought to throw light on the relations between Alans and Ossetes from several directions. He examines, first of all, the place-names in the territory inhabited by the Balkars and the Karachay, and discovers numerous Ossetian elements in them; on the basis of these elements he comes to the conclusion that the territory was once inhabited by people who spoke Ossetian, or, more precisely, the western dialect of that language. He points out, on the other hand, that, according to the testimony of medieval sources, the Balkar and Karachay territories used to be inhabited by Alans, and that as a matter of fact, the Karachay are to this day called alani by the Mingrels. These facts, in Abaev's view, can be explained only by supposing that historical contacts between Alans and Mingrels must have existed during the Middle Ages. The inscription of Zelenchuk, found at a site north of the present Karachay territory, is regarded as being Ossetian by Abaev who, on this point, follows Miller's view. Abaev also discusses in detail linguistic contacts between Hungarians and Ossetes. He has no doubts that there is a stratum in the Hungarian and the Ossetian vocabulary common to both languages, this leads him to the conclusion that at a definite historical period there must have been two contiguous linguistic communities; the descendants of one of these communities are the Hungarians of to-day, the descendants of one of these communities are the Hungarians of to-day, the descendants of the other are the present Ossetes. Thus, taking the historical continuity of Alans and Ossetes as his basis, Abaev thinks that the people who enriched Hungarian with Ossetian elements, could only have been the Alans. He tries to illumine the problem of historical contacts between Alans and Ossetes, also by examining Alanic person's names. Abaev points out that the Alanic name Ma-ta-rh-sha, known to us from a Chinese record, has an exact equivalent in the present Ossetian name Matärsa, while the mane A-da-chi has a corresponding Alanic form Addac in the fifth century. Finally, Abaev discusses in detail the interpretation and significance of the Alanic formulae of salutation preserved in Tzetzes, from the angle of Alanic — Ossetian relations. He demonstrates that the Alanic words found in Tzetzes show close affinity to present Digorian (Western Ossetian) forms. Nevertheless, in summing up the results of his investigations, Abaev expresses his conviction that "a great many of those peculiarities, which nowadays separate the Ironian (= Eastern Ossetian) dialect from the Digorian, did not exist at that time (in the eight century), and the (linguistic) facts established by Tzetzes reflect, not some specific "Digorian" forms, but the "average" Alanic forms of that age." [21]

Abaev's work has, in many details, greatly contributed to research intended to clarify relations between Alans and Ossetes. But on the whole, Abaev's point of view is closely related to Miller's attitude which he had rejected so sharply, in principle. The fact is that Abaev denies the existence of the present dialects in medieval Ossetian, i. e. regards Alanic as a uniform language, and admits the theory of a direct

Additional Notes

To p. 62. On the Ossetian dialects cf. V. I. Abaev: . I. 357 foll., G. Akhvlediani: . I. 60 foll., M. I. Isaev: . Moscow 1966. 5—112 and  14 (1965) 140 foll.

To p. 62. The most important works of V. I. Abaev are now joined in his valuable book . I.

To p. 62. My review of the results and theories of my esteemed friend Professor V. I. Abaev is very incomplete. The whole richness of his life-work is now to be found in his . I. and . I. Moscow—Leningrad 1958.

20. Alanica. 1935, , 881 foll.

21. With regard to Abaev's conclusions see also D. Gerhardt's detailed review, amounting practically to a translation, in ZDMG XCIII (1939), 33 foll.


Alanic—Ossetian historical continuity: these features of his attitude are hardly influenced by the circumstance that he does not regard the Alans simply as the "forebears" of the Ossetes, nor the Ossetes as the "descendants" of the Alans. Abaev's whole view rests fundamentally on the family-tree theory, as did that of Miller: in accordance with this basic conviction Abaev would derive the Ossetian dialects of to-day from a uniform medieval Alanic language. This view reflects, no doubt, the conviction that if we reverse the flow of time, we meet with increasingly uniform states of language. It is enough to give two examples, in order to show to what extent this conception influences Abaev's work. In analysing the Alanic word χaς, he is only anxious to stress that the word stands nearer to the Digorian form χwarz than to the Ironian form χorz. In Abaev's opinion, it is usually the Digorian dialect that represents the older phonemic stage; consequently, if the phonemic form of the Alanic word is closer to the Digorian form, this would prove clearly, on the one hand, that there is direct historical connection between Alanic and Ossetian, and, on the other, that the Ironian phonemic form must have been, formerly, the same. Meanwhile. Abaev fails to notice that it is impossible to deduce the present Digorian and Ironian forms from Alanic χaς (: χas, χ, χaz, etc.), so that this word, instead of lending support to, actually refutes the theory of direct historical connections between Alans and Ossetes. Similarly, in connection with the Alanic word χσινα the only thing Abaev notices is the presence of the final phoneme -a which appears also in the Digorian form äχsinä (in contrast to Ironian ’χsin). In this case both the Digorian and the Ironian forms may be derived, without any special difficulty, from Alanic χσινα: but the Hungarian word asszony (Old Hungarian achscin, : aχsin), borrowed from Alanic before the tenth century, definitely points to a form, aχsin. It follows from this that, as early as the tenth century, two forms, aχsin and χsina, must have been in living use, i. e. the present dialectal differences in Ossetian must have already existed then. [22]

In contrast to Miller's view, according to whom Alanic-Ossetian was in direct historical connection with the language of the Scythians and Sarmatians, Andreas had previously expounded his theory that the Alans were not Sarmatians, but later immigrants into Southern Russia from their Eastern Iranian home in Xwārizm. [23] Andreas' arguments, unfortunately, did not appear in print, so that his conception had no serious response for a long time. Meanwhile Charpentier, too, started advocating the theory of the Eastern descent of Alans and Ossetes, deriving his arguments from historical sources. He conjectured that the original tribal name of the Alans was as- or os-, so that the Alans may be regarded as being identical with the  who, according to Strabo, had conquered Bactria, with the Asiani of Trogus Pompeius, and the Wu-sun of Chinese sources. [24] Charpentier's conclusions would have had, of course, far-reaching linguistic consequences if only they could have been verified. But the necessary linguistic material was missing at the time. The eastern linguistic contacts of Alanic—Ossetian could be tackled, with any hope of success, only after Chorasmian texts had come to light in considerable quantities, i. e. when it became possible to form some idea of the language of Xwārizm, the territory from which Andreas had long ago sought to derive the Alans and Ossetes.

Additional Notes

To p. 63. Against the confrontation of Alanic χsina with Old Hungarian aχsin Abaev argues with reference to the momentary, unstable character of the Ossetian prothetic vowel ä. It must be noted, however, that from the view-point of Hungarian aχsin > asszony the prothetic a in the Alanic prototype of this loan-word must have been a very stable vowel because otherwise it would not have prevailed against the force of vowel harmony. Thus my confrontation of Alanic χsina with Old Hungarin aχsin becomes even more valid.

22. Abaev's latest book: . T. I.  1949, has so far been inaccessible to me.

23. See A. Christensen, Die Iranier: Handbuch der Altertumswissenschaft. III, Abt. I. Teil, III, Bd. III. Abschn. 1. Lief. München 1933, 249, note 2. Andreas himself gave a brief outline of his position in Verhandl. d. XIII. Intern. Orientalisten-Kongresses. Leiden 1904. 103.

24. ZDMG LXXI (1917), 357 foll. Of the tribes identified by Charpentier we have to exclude, in any case, the Wu-sun, for historical and geographical reasons; see G. Haloun, ZDMG XCI (1937). 252.


It was Zeki Validi who first succeeded in discovering Chorasmian texts in any quantity, and who found a passage in Bīrunī (in the Introduction to the Taḥdīd nihāyāt al-amākin) which seems to be of decisive importance in forming a judgment about the language of the Alans. According to Validi, the passage in Bīrunī informs us that the "Alans and Ās had formerly lived, together with the Pechenegs, around the lower reaches of the Amu-Darya (the Uzboy), and later, after the river had changed its course, they migrated to the coast of the Sea of the Khazars"; Birūnī also tells us that "the language of these Alans is a compound of Chorasmian and Pecheneg-Turkish". Validi takes this to mean that the Chorasmians spoke an Iranian language related to Ossetian; he thinks it likely, at the same time, that the language of these Alans, who had migrated to the land of the Khazars, must have differed in some measure from the language of the Caucasian Ossetes. [25]

It was Henning who first subjected to linguistic scrutiny the Chorasmian texts discovered by Validi; he came to the conclusion that, although the Chorasmian language shares many important characteristics with Ossetian, nevertheless, on the whole it is nearer to Sogdian, while it also has a number of characteristic features found neither in Sogdian nor in Ossetian. The features shared with Ossetian consist, according to Henning, chiefly of the phonemic changes š > s and č > c, though the change from š > s was not entirely completed in Chorasmian. [26]

Thus the scrutiny of Chorasmian texts has for the time being failed to supply linguistic facts that might be regarded as a decisive proof of the theory affirming the Chorasmian origin of the Alans. This circumstance obviously influenced Validi's mind when he came to the conclusion that the language of the Alans, who had migrated to the land of the Khazars, must have been somewhat different from that of the Caucasian Ossetes. This is, naturally, equivalent to admitting that the Chorasmian origin of the Alans—Ossetes (a conjecture based on considerations of history) cannot be proved as a linguistic proposition.

These negative linguistic conclusions, which contradict the evidence of historical sources, were, naturally, far from reassuring to those advocating the eastern origin of the Ossetes; hence several new attempts were made recently to try and prove the close contact of Ossetian with the languages of North-Eastern Iran or its eastern origin. Among these attempts let us first consider Freiman's works. He discovered a considerable quantity of fresh Chorasmian linguistic material, and in elaborating it touched several times on the question of the relation between Ossetian and Chorasmian. Freiman's investigations have established that correspondences between Ossetian and Chorasmian are not restricted to the phonemic changes š > s and č > c, pointed out by Henning, but extend to a number of phenomena of different kinds. Thus Freiman has shown that the phonemic change -ti > -ci is found both in Ossetian and in Chorasmian: see e. g. Chorasmian akic '(he) makes' ~ Ossetian känənc '(they) make'; in some cases the Old Iranian group of phonemes -ϑr- has similar corresponding forms in both languages, e. g. Chorasmian arcivak 'third' ~ Ossetian ärtä 'three'; Old Iranian initial h- has disappeared in many cases both from Ossetian and Chorasmian, e. g. Chorasmian iβdac 'seventy' ~ Ossetian ävdai 'seventy' ~ Old Iranian haptāti; the plural suffix - characteristic of Ossetian is found also in Chorasmian, e. g. niγōsic 'listeners', nikanc 'stakes, posts', sparc 'shields'. [27] Freiman attributes very great importance to these correspondences when pronouncing judgment on the

Additional Notes

To p. 64. On the position of the Chorasmian cf. W. B. Henning: Handbuch der Orientalistik. I. Abt. IV. Bd. l. Abschn. Iranistik. I. Abschn. Linguistik. Leiden—Köln 1958 109 foll. Henning judges about the common features of Chorasmian and Ossetian rather negatively.

To p. 64. As W. B. Henning, The Khwarezmian Language. Zeki Validi Togan'a Armağan. Istanbul 1955. 10 has shown, the Chorasmian plural suffix -c goes back in fact to -k.

25. See ZDMG XC (1936), *26* foll, and also Ibn Fadlān's Reisebericht, Leipzig 1939, 14, 125 foll., 137. [Add. note: Cf. also V. Minorsky, Ḥudūd al-‘Ālam, London 1937, 481.]

26. ZDMG XC (1936), *30* foll.

27. See , CB IV (1947), 157 foll., CB V (1948), 191 foll., CB VI (1949). 63 foll.


origin and place of the Ossetian language. He states emphatically that "the transference to the West of our knowledge concerning the linguistic Middle Ages of Eastern Iran (this is Freiman's description of the discovery and elaboration of the Chorasmian linguistic material) has made it possible to lay a firm foundation for those linguistic bridges which connect more closely the Chorasmian language with the language of the Alan—Ās, i. e. with the language of the Ossetes, those emigrants who had their homes in Xwārizm". [28] In one instance Freiman makes the attempt to trace back the connections of the Ossetian and Chorasmian or Saka languages, as far as the fifth century B. C.: he tries to explain the name Skunxa, the Saka chieftain defeated by Darius, from the Ossetian verb skuänxun 'to be different, to be/get distinguished'. [29]

Tolstov has called attention to another interesting proof of the Chorasmian origin of the Alans—Ossetes. He pointed out that one of the Turkmen tribes of South-Eastern Turkmenia bears the name Alan, a name which denotes also one of the subsidiary tribes of the Salïrs. According to Tolstov, the Turkmen tribe Alan differs in a number of ethnographic peculiarities from the surrounding Salïrs: one may observe among them, for instance, a strong tendency toward tribal endogamy and marriage within the clan; they wear white clothing, etc. It is especially noteworthy that a tradition has been preserved among them, according to which they migrated to their present habitation from the Mangïshlak Peninsula where, they say, there used to be "a large fortress known by the name of Alan". The interesting point is that there exist, in fact, ruins of a fortress known as Alan-kala ("Alan fortress") on the north-western borders of Xwārizm, between the Sea of Aral and the Mangïshlak Peninsula. So there can be no doubt that the tradition of the Alan Turkmen tribe has a historical value, and that we may regard this tribe as Turkicized descendants of the Alans who used to live on the territory of Xwārizm and on the plateau of Ust-Urt. [30]

It was Tolstov, again, who pointed out that the name of one of the Chorasmian rulers appearing on his coins as wrϑwmχ, while in Bīrunī it figures in the form ’rϑmwχ, bears a close resemblance to the name of Uruzmäg, a well-known hero in the Nart sagas of the Ossetes. [31] This correspondence — if it can be linguistically verified — supplies another interesting datum for the historical contacts between Alans—Ossetes and Chorasmians. We may establish, at all events, that the passage in Bīrunī and the reading of the Chorasmian coins give two different forms of the name: 1. warϑumaχ and 2. arϑamuχ. But the same duality appears also in Ossetian as, beside Uruzmäg, there also occur the forms Wäräzmäg, Oräzmäg, and Wərəzmäg. [32] On the basis of these and the Abadzech form Urzames we may suppose the existence of an earlier form *Warzəmag ~ *Warzumag which is quite close to the form Warϑumaχ of the Chorasmian ruler's name.

There is no doubt that Freiman's observations and Tolstov's data have brought forward a lot of important new material to the question of Alanic—Ossetian history and language. But we must not ignore the fact that, while Freiman's researches have considerably increased the number of linguistic correspondences between Ossetian and Chorasmian, they have also revealed more fully that Chorasmian stands much

Additional Notes

To p. 65. On Chorasmian ’rθmwχ cf. now V. A. Livshits: Acta Ant. Hung. 16 (1968) 442 who shows that this name does not occur on the coins. On Uruzmäg cf. V. I. Abaev:  10 (1945) 25 foll. (*Varāz-man), . I. 92, E. Benveniste: Etudes sur la langue ossète. 129 (*Avarazmaka-).

28. See , VII (1948), 238 foll.

29. Ibid 239.

30. See  1948, I, 197. Similar data with regard to the Alans near the Sea of Aral, as e. g. Firdusi's Diž-i Alānān and the place-name Qïzïl-Alan in the Turkmen steppes, have been earlier pointed out by Marquart, Über das Volkstum der Komanen, AGGW XIII, Berlin 1914, 106 foll, and by Minorsky, Ḥudūd al-‘Ālam, London 1937, 481.

31. See  1948, 189,  1948, 161, foll.

32. See  V (1935), 281.


closer to Sogdian than to Alanic—Ossetian. For this reason we need not be surprised that some scholars, e. g. Altheim, continue to regard the passage in Bīrunī about the language of the Alans and Chorasmians just as problematical as before. According to Altheirn, Bīrunī could certainly not mean that the Chorasmian and Ossetian languages were especially close to each other with regard to their origin: the meaning of the passage is that the Alans or Ās took over certain linguistic peculiarities from the Chorasmians, in whose neighbourhood they once lived, and that the same applies also to the Pechenegs. [33] For the rest, Altheim accepts the identity of the present Ossetians with the medieval Ās and the ancient  the conquerors of Bactria, i. e. he accepts the thesis of the eastern origin of the Ossetes. [34] His attempt, however, to interpret the passage in Bīrunī in the light of late historical contacts between Chorasmians and Ossetes, instead of assuming an identity of origin or linguistic community between these two pwoples, must be, therefore, ascribed to a negative estimate of the linguistic connections between Alanic—Ossetian and Chorasmian.

Parallel with the linguistic research on the relations of Ossetian and Chorasmian there also emerged several historical combinations which tried to solve the origin of the Ossetians and the Alans in the direction indicated by Charpentier. One of these combinations is Vernadsky's. He has renewed the conjecture about the supposed identity of the Wu-sun and the , as well as the Asiani, the As, and the Ossetes. He has, moreover, introduced new elements into this combination by trying to prove that the names Anti and Yen-ts’ai belong to the same group of peoples' names. [35] But these combinations of Vernadsky's raise very serious historical and linguistic difficulties. [36]

Maenchen-Helfen also follows in Charpentier's footsteps with regard to the origin of the Ossetes and the Ās, [37] but by utilizing the results of recent investigations he is able to set this problem into a much wider framework. Under the influence of Haloun's arguments, Maenchen-Helfen rejects the identification of the Wu-sun and the Asiani, and proposes a new, wider combination in its stead. He tries to prove that the name Ārśi used by the Tokharians about themselves is identical with Pliny's Arsi, Ptolemy's  as well as with the Aorsi who came to be called Alans later on. These peoples or peoples' names, to which he adds the al(-l)ārisiya mentioned in Mas‘ūdī, are, in his view, identical with Ās, the old name of the Ossetes and its different varieties. All these peoples are, at the same time, Tokharians, i. e. the Yüeh-chih of the Chinese, since Ārśi is the name used by the Tokharians for themselves. In Maenchen-Helfen's opinion the name Tokhar, itself, is found among the Ossetes in the tribal name Digor. Maenchen-Helfen, himself, must have felt that these identifications of peoples and peoples' names raise a host of historical difficulties. For this reason he tried to render them more probable by assuming the presence of a number of historical layers. According to his account, the tribal name of the Yüeh-chih was Togar, while their ruling group bore the name of Kusha (transcribed as Yüeh-chih by the Chinese). This peoples came under the rule of the Sacae who called themselves Ārśi (= Aorsi, Arsi, , Asiani, Ās, etc.). The people, formed as the result of this Togar-Ārśi stratification, was later divided into several groups. One

Additional Notes

To p. 66. Cf. also F. Altheim: Aus Spätantike und Christentum. Tübingen 1951. 59 foll, and Geschichte der Hunnen. I. Berlin 1959. 57 foll.

33. F. Altheim, Literatur und Gesellschaft im ausgehenden Altertum, Halle/Saale 1950, II, 210.

34. See Der Hellenismus in Mittelasien: Saeculum I (1950), 281.

35. G. Vernadsky, Ancient Russia,3 New Haven 1946, 82 foll., Byzantion XVI (1942—44), 81 fol.

36. See my remarks in RHC N. S. V (1947), 230 foll.

37. JAOS LXV(1945), 71 foll. O. Maenchen-Helfen himself refers to Charpentier but he exaggerates in connecting the identification of Ārśi-Asiani with Charpentier (79), since the word Ārśi was introduced into the Tokharian controversy only by Sieg SBAW 1918, 560 foll.


group migrated towards the West, and became the ancestors of the As-Digūr among the present Ossetes. Maenchen-Helfen distinguishes, morever, the Alans from the Ās. The upshot of these identifications is that, while the Tokharian problem becomes over-simplified, the formation of the Ossetes turns out to be the result of a very complex ethnical stratification.

There is no doubt that, even with the assumption of these historical strata, Maenchen-Helfen's conclusions contain many elements that are hypothetical or entirely unsupported. His attempt, however, to explain the formation of the present Ossetian people as the result of repeated ethnical stratifications, in contrast to former conjectures deserves close attention, in any case.

H. W. Bailey's recent investigations in the study of the origin of Ossetian vocabulary have a very important bearing on the contact of Ossetian with the Eastern Iranian languages as well as on the eastern origin of the Ossetes. Since the studies of Hübschmann (Etymologie und Lautlehre der ossetischen Sprache) and Miller, Bailey's works may be regarded as the most important step forward in the study of the origin of the Ossetian vocabulary. Bailey does not connect the Wu-sun with the Asiani; he even dismisses the nameĀrśi which he regards simply as the Tokharian equivalent of the Northwestern Prākrit form of the Sanskrit word ārya- 'beggar monk'. Thus he ultimately identifies the old Ās and the present Ossetes only with the . At the same time, he derives the name Ās, Ossetian Asi ~ Assi from an earlier form *ārsya-, and connects this with the al-(l)ārisiya found in Mas‘ūdī as well as with the names Arsi and . Thus Bailey regards the Ossetes as the descendants of the , an Eastern Iranian tribe which conquered Bactria: he attempts to support this view with the results of his study in the field of the Ossetian vocabulary. He tries to prove the presence in Ossetian of a considerable number of words, the exact equivalent of which can be demonstrated only in Sogdian and Saka. In Bailey's view, these correspondences indicate that the ancestors of the Ās were in close contact with the Chorasmians, Sogdians, and the forebears of the Afghans. This symbiosis is put by Bailey to the third century B. C. since the Iranian names in the Greek inscriptions of South Russia, and the earliest linguistic remains of the Sogdians, (both types going back to the second century A. D.) reveal, in Bailey's opinion, clearly defined linguistic individuality, so that the state of symbiosis must have existed several centuries before. [38] This train of thought shows also that, during the period of symbiosis of the Ās, Sogdians, Chorasmians, etc., Bailey assumes the linguistic community of their respective languages, otherwise he might just as well have assumed the existence of a state of symbiosis at a later period when these tongues developed into fully-fledged separate languages. Thus, it would seem that, ultimately. Bailey sees the relation of these languages to one another from the angle of the family-tree theory.

Bailey's works have considerably enriched our knowledge concerning the Eastern Iranian contacts of the Ossetian language, in general, and the Ossetian vocabulary, in particular. But while stressing this, we cannot fail to remark that his conclusions cannot, in all respects, be regarded as final, either from the historical or the linguistic point of view. First of all, there is no need whatever to assume linguistic unity, for a period, when peoples speaking different languages are living together. We have seen above that the Ās and thr Chorasmians were living together as late as the tenth century A. D. — yet there is no question of a linguistic unity between

Additional Notes

To p. 67. During the last two decades H. W. Bailey gave important contributions to the historical analysis of the Ossetian vocabulary in almost everyone of his papers.

38. See H. W. Bailey, TPhS 1945, 1 foll., TPhS 1946, 202 foll. TPhS 1947, 142 foll., 150 foll., BSOAS XIII (1949-50), 135.


Ossetian and Chorasmian. So there is no inevitability, either, in Bailey's deduction, according to which Ossetian must have been living together or at least have been in contact with other languages of Eastern Iran about the third century B. C. From the methodological angle, too. Bailey's procedure of trying to determine the relation of Ossetian to the Eastern Iranian languages, on the basis of vocabulary, is open to objection, especially if we have to count in Ossetian with a complex Eastern Iranian stratification. Thus it is clear that the problems raised and discussed by Bailey are still waiting to be examined from a number of different angles.

After this survey of recent research on the position of the Ossetian language, we see clearly those major groups of problems which it is necessary to solve if we wish to attain a certain degree of certitude with regard to the Eastern Iranian connections of Ossetian, or the problem of the North Iranian group of languages as a whole. These groups of problems may be summed up as follows:

1. The relation of Ossetian to the ancient Iranian languages of South Russia. The clarification of this problem is indispensable if we want to see clearly the relation of Ossetes, Alans, Sarmatians, and Scythians.

2. Within the above group of problems the question of plural formation with  requires a separate examination since it has always been a pivotal question in research and the available material is considerable. In the eyes of the majority of scholars this method of forming the plural is one of the decisive proofs for the close connection of Scythian—Sarmatian—Alanic—Ossetian on the one hand, and of the Eastern Iranian languages, on the other. The question, however, is whether this plural suffix really existed in Scythian, and whether one is justified in regarding this morphological peculiarity of the language as a dialectological criterion.

3. It is necessary to clarify the mutual relations of Alanic and Ossetian. This work requires, of course, a thorough re-examination and re-valuation of the linguistic remains of the Alans.

4. The solution of the same problem also requires the re-examination and re-valution of the Alanic loan-words in Hungarian. As we have seen above, Abaev ascribed a very important role to these loan-words in clearing up the relation of Alanic and Ossetian. Their testimony was regarded as decisive by Sköld, too, in the question of Ossetian dialects.

5. The relation of Ossetian to the Eastern Iranian languages. The discovery of the Chorasmian texts, the results of historical research, as well as the works of Henning, Freiman, and Bailey on the subject, have made the clarification of this problem one of the most pressing tasks of Ossetian linguists.

6. The stratification of the Iranian elements in the Ossetian vocabulary. This question was raised by the possibility that the Ossetian people were formed by various Iranian tribes being superimposed, one upon the other. The existence of such a possibility was clearly demonstrated by Maenchen-Helfen's results, even if the latter require substantial corrections in many respects. Moreover, if we have to count with different ethnical strata in the case of the Ossetian people, this must find a reflection in their vocabulary, too. Thus, this question is one of the most exciting tasks of future research.

Of these groups of problems, we are going to discuss in this essay the relations of the ancient Iranian languages of Southern Russia to one another, and to Ossetian.

[Previous] [Next]
[Back to Index]