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


3. The Sarmatian Dialects of the North Pontic Region

Thus, in examining the Iranian names preserved in the Greek inscriptions of South Russia, we have to clear up the question as to whether these names really reflect a homogeneous language, and whether this language may really be regarded as the predecessor of Ossetian. The question is, what criteria have we to employ in attempting to solve this problem. Miller's proofs, as we have seen above, are partly of phonemic, partly of morphological character. As to the two morphological proofs, i. e. the presence of the suffix -äg and -gin in the names of the inscriptions, undue importance need not be attached to them. Miller himself has pointed out that the suffix -äg is found also in Persian, in Baluchi, and in Afghan. [72] One may

72. Die Sprache der Osseten, 89.


add that it has since been discovered in Sogdian, too; cp. e. g.šm’r’k 'qui a des scrupules, qui hésite', from the verb šm’r-; np’yš’k 'qui ecrit' from the verb np’yš-, etc. [73] The same applies also to the suffix -gin, equivalents of which have been found by Benveniste in Persian and Sogdian. [74] Thus these morphological elements cannot be utilized to prove the existence of nearer genealogical connections between the language of the Iranian tribes in South Russia and Ossetian. In order to clarify the relation between the Iranian names in the Pontic Greek inscriptions and Ossetian, we shall have to rely on phonemic criteria. This implies, no doubt, a certain degree of one-sidedness; but this one-sidedness is the direct consequence of the linguistic material at our disposal. Moreover, if we have to count with the existence of several dialects among the Iranian tribes of South Russia, these differences may be best determined on the basis of phonemic criteria. The differences might become even sharper if we were able to define the phonemic system of the languages or dialects used by the various Iranian tribes of South Russia; but the one-sided linguistic material, preserved only in Greek transcription, does not make this feasible. Hence in the rest of this essay we shall have to limit ourselves to the demonstration and systematization of phonemic differences. [75]

If we examine the Iranian names preserved in Greek inscriptions as well as the Sarmatian names figuring in the works of classical antiquity, we shall find that names dating from roughly the same period show different developments for certain Old Iranian phonemes or groups of phonemes. This fact is of outstanding importance since the most obvious explanation of the phenomenon is that names, showing different lines of phonemic development but dating from the same period, derive from different dialects. This, in turn, leads us to the conclusion that the Iranian (Sarmatian) tribes in South Russia spoke dialects or languages more or less different from one another. These languages and dialects were, of course, possibly more closely connected among themselves than with any other Iranian language. The most characteristic cases of this type are the following:


The Old Iranian initial group of phonemes *ar-y- has four different developments: 1. ar-(y)-, 2. al-, 3. ir-, 4. il-.

Old Iranian *ar-y-  > ar-(y)-.

Arii 'name of a tribe in South Russia' in Epiphanius (4th cent. A. D.), De lapide ligyrio V. IV. p. 190 Dind. (, I. 712): < Old Iranian *arya- ~ Avestan airya- (Vasmer, Die Iranier in Südrußland. 33).

Arraei Sarmatae, quos Areatas voncant: Pliny, Nat. Hist. IV 41 (1st c. A. D.). Thus Pliny gives two names for this Sarmatian tribe living south of the Danube. In the ending of the second name we recognize the plural suffix -t, -, found also in Ossetian, Sogdian, and Yagnobi. The remaining first part of the name (Area-) may be compared again with the Old Iranian name *arya-. The difference that appears in the second syllable of the two forms (-ea- ~ -ya-) may be explained in two different ways. First, taking into consideration the fact that in the transmission of Pliny's text the names of peoples are to a varying degree, and sometimes hopelessly, corrupt,

Additional Notes

To p. 77. L. Zgusta, Die Personennamen griechischer Städte der nördlichen Schwarzmeerküste. 263 made the surprising statement that we do not know anything about the ethnical qualification of the Arraei Sarmatae. Does the qualification Sarmatae not say him really anything?

73. See Benveniste, Essai de grammaire sogdienne, II, 55.

74. Essai de grammaire sogdienne, II. 97.

75. On the classification of dialects sec Gy. Laziczius, A magyar nyelvjárások, Budapest 1936, 44 foll. For the criticism of earlier methods see , 228 foll.


we may easily suppose that Area-tas is a corruption of an earlier form *Aria-tas. This form would correspond precisely to a possible Sarmatian form *arya-ta. The second, more likely possibility is that the spelling Area- reflects a form ara- which is simply a variation of the word arya-. There is a parallel among the names preserved in the inscriptions: the name  known from two inscriptions of Tanais, the nearest possible transcription, in Greek letters, of an Iranian form syāvak ( ~ Avestan syāva- 'schwarz') appears in the variation  in an inscription of Panticapaeum (see Vasmer, op. cit., 51 foll., Schaeder, Iranica, 51). This latter datum makes it clear that the spelling Area- may be also a transcription of the word arya- or one of its variations (? arəya- or ara-). If we take into consideration that the form Areatas contains an Iranian morphological element (the plural suffix -t) , it seems very likely that this name was used by the Sarmatian tribe in question, to denote itself. It is possible that this name is somehow connected with the one discussed above. The most natural assumption would be that this, in its turn, is a Latinized (or, if Pliny took his datum from a Greek source, Grecized) form of the word arya-. In this case, however, we must regard it as a corruption of the form *Arii or *Arei. To show that such a distortion might easily occur in the transmission of Pliny's text, it is enough to quote the manuscript variants of the name Pangaei: pangaei, pangei, paegiae, pegei, peie.

Diod. Sic. XX. 22, 4 < *arya-farnah- (Vasmer, loc. cit.).

Old Iranian *ar-y- > al-.

'Iranian tribe of Southern Russia' Flav. Joseph. VII 7, 4. The name of the Alans is known from other sources, too: e. g. Chinese sources: A-lan (Hou-Han-snu, 118, Wei-lüe, fragm. 22. see Junge, Saka-Studien, 77); Latin authors: Alani and Halani; Arabic and Persian sources: al-Lan; recently the name has been discovered also in Mingrel, one of the Caucasian languages: alani-koi 'Alanian mani.e. strongbravehero, champion' (see  OOH 1935, 883). All data in the sources point to the form alan. The name alan admits of two acceptable interpretations. According to Andreas (cf. H. Jacobsohn, Arier und Ugrofinnen. 234) and Sköld (Die ossetischen Lehnwörter im Ungarischen, 68) it goes back to the plural genitive form aryānām of the Old Iranian word arya-; hence, morphologically, it is ultimately identical with the Persian name Ērān, Īrān. According to the other interpretation, also the name alan goes back to the Old Iranian word arya- 'Arier' (more correctly, perhaps, to ārya-, see Tedesco, ZII II (1923) 46); not, however, to the genitive plural but to the form ryna- formed with the suffix -na-. From the angle of phonetics, assuming the development -ry- > -l-, [76] both interpretations are irreproachable; yet, owing to semantic considerations, the second explanation must be preferred. The name ĒrānĪrān is the name of a country and probably developed from the construction rynām χ
šaϑramērān šahr 'das Reich der Arier'. All our data, however, which refer to the name alan (with the exception of the Chinese sources, the testimony of which, in this respect, is by no means decisive) agree in indicating that this name was the name of a people. The word rya- and its derivation ryna- were used to denote a tribe or a tribal federation. An enlightening parallel to this use may be found in the tribal names discussed above (Arii and Arraei ~ Areatae) as well as in the name the Eastern Ossetes use about themselves, viz. the name ir 'Ironer (Ostosseten)' which may also be traced back to the Old Iranian form

Additional Notes

To p. 78. The problem of the name Alani was often discussed in the meantime. At first I would add A. Freiman: RO 3 (1925) 158 foll, (alān < aryaanām ir < ēr < err < arya-) to my references. Cf. further V. 1. Abaev: . I. 156, 245 foll., F. Altheim: Aus Spätantike und Christentum. 63, 168 und Geschichte der Hunnen. I. 58 foll., L. Zgusta: Die Personennamen griechischer Städte der nördlichen Schwarzmeerküste. 264, I. Gershevitch: BSOAS 17 (1955) 486, V. I. Abaev: . I. 47 foll., 545 foll., H. W. Bailey: TrPhS 1959. 98 foll., G. Akhvlediani: . 211 foll. As to the recent attempts to elucidate the origin of the name Alani, I restrict myself to a few short remarks here.

1. A form *ārya- cannot be evidenced in Iranian languages, accordingly we can only suppose an Old Iranian form *arya- as starting point. Thus the form allon cannot be a genuine development from *aryāna- in Ossetian.

2. Alan is a name which is never applied by the Ossetes to themselves. This fact speaks against the Ossetian origin of this term.

3. The development -ry- > -l-cannot be proved in Ossetian reassuringly, because all reliable instances quoted by Abaev and others only show -ri- > -l-.

4. The Eastern Ossetes use the name ir, iron to denote themselves, consequently this name must be genuine and cannot be regarded as of Caucasian origin. Because the Western Ossetes do not denote themselves with the term irä, this form may be borrowed by them from the Eastern Ossetes.

5. The name asi is used by the Ossetes as the denomination of Balkaria where the place-names prove the former existence of a population speaking an Ossetian dialect.

From all these facts it follows that the problem of the historical relations between Alani, Ās and Ossetes is a very complicated one. I hope to come back to this question.

76. Andreas, quoted by Gauthiot, Essai de grammaire sogdienne, I, III,; Jacobsohn, Arier und Ugrufinnen, 234.


rya-. The form ryna- itself is found in the Avesta: airyana- 'arisch', airyanəmχvarənō, airyanəmvaēō (Bartholomae, AirWb. 198). Recently, the from ryna- has also been traced in Sogdian in the construction rynāmvaiah: ’rynwyjn: aryān vēžan (see Henning, ZDMG XC [1936], 5). Since, however, the above word does not occur otherwise in Sogdian, we may possibly have to do with a borrowing from Persian in this case. We have to point out. however, that none of these data testifying to the existence of the word ryna- are used to denote the name of a people, or the name of a definite tribe. For this reason, from the angle of deriving the word alan as the name of a people from the form *ryna-, decisive importance attaches to the fact that in Ossetian we actually find the word rya-, with the suffix -na- appended, used to denote an ethnical unit: iron 'Ironer (Ost-Ossete), ost-ossetisch' < Old Iranian *ryna-.

The derivation of the name alan from the form *ryna- was earlier regarded with some scepticism by Vasmer; [77] later, however, he came to accept this interpretation [78]. In the eyes of those advocating the identity of Alans and Ossetes, the most surprising feature in this derivation of the name alan from the form ryna-, was the need to suppose the presence of the phonetic change -ry- > -l-, a change that cannot be shown to have taken place in Ossetian. In Ossetian the Old Iranian initial group of phonemes *ry- does not develop into al-, as one would expect on the basis of the name alan and the supposed identity of Alans and Ossetes: it develops into īr- or īr- (cf. Eastern Ossetian īr 'Ironer (Ostosseten)' Western Ossetian īrä 'Ossetianrace, breedOssetian people' see Munkácsi, KSz XXI (1932), 86, < Old Iranian *rya-; Eastern Ossetian iron 'Ironer (Ost-Ossete), ost-ossetisch' < Old Iranian *ryna- [79]). Hence the name  can in no wise be regarded as Ossetian because its phonemic form cannot be brought into line with the phonemic correspondences between Old Iranian and Ossetian. Nor does the situation change if we reject the view discussed above, and suppose that the name  goes back only indirectly to *arya-na- [80], because, even assuming this, we cannot explain the presence of the initial a- from Ossetian. The fact that the name Alan cannot be interpreted from Ossetian has a decisive importance on our judgment about the relations of Alans and Ossetes. It would be an obvious assumption, of course, that the name Alan was not applied by the Alans to themselves. This possibility, to which there can be no objection in principle, is, however, excluded by the fact that the change -ry- > -l- may be demonstrated also in the case of another name, precisely on the linguistic territory of the Alans (see the following item).

Phanagoria, Latyshev, IOSPE II, 363 (307 A. D.): Vasmer, in Die Iranier in Südrußland, 31 explains this form as deriving from Old Iranian *aryašaϑra- through dissimilation, in RLV XII 224 as deriving from the same form, through popular etymology, under the influence of . But the names of the inscriptions do not furnish any examples for such dissimilation (cp.  Vasmer, Die Iranier in Südrußland, 33): as to connecting this word with the Greek name , by way of popular etymology, this could have occurred only after the change ary- > al- had taken place: hence it seems certain that, as in the case of the people's name , here, too, we have to do with the change ary- > al-, having the force of a phonetic law. Old Iranian *aryašaϑra- developed into *alašarϑ and this form may have turned, being connected in popular etymology

Additional Notes

To p. 79. As a parallel to  reflecting the earlier stage of phonemic development, the name  (Eltegen, 2nd half of the 1st — 1st half of the 2nd centuries A. D., CIRB No. 931) can be quoted. The assumption of Zgusta (Die Personennamen griechischer Städte der nördlichen Schwarzmeerküste. 64) that  might be a clerical error for  is a mere invention.

77. Die Iranier in Südrußland, 31.

78. RLV XII, 242.

79. Hübschmann, Etymologie und Lautlehre der ossetischen Sprache, 41.

80. Vasmer, loc. cit.


with the Greek word  or  into . Another possibility is that the form  does not owe anything to Greek popular etymology but simply reflects a form *aläχsarϑ. One is justified in supposing the existence of such a form, on the analogy of parallel forms like  < Old Iranian *kainaχšaϑra- and : kēnäχsarϑ.

Phanagoria, Latyshev, IOSPE II, 389. If this, in fact, is the correct form of the name (it does not seem quite certain), then the word reflects an Iranian form maldaγ which, in turn, may have developed from an Old Iranian form *marya-taka-, under the operation of the change -ry- > -l-. The element marya- in this word may be regarded as the exact equivalent of the Old Indian word marya- 'Jüngling, flotter junger Mann', while the second part -taka- may stand for a group of suffixes. This latter can probably be divided into the elements -(a)t-a-ka-. With regard to the suffix -at-, -t-, cp. Avestan brvat- 'Augenbraue' ~ Old Indian bhrū- 'Augenbraue' and Avestan hu-zāmit- 'leicht gebärend' ~ Avestan hu-zāmay- 'leichte Gehurt' (H. Reichelt, Awestisches Elementarbuch, 152); as to the juxtaposition of the suffixes -t-and -t-a-, cp. Old. Indian harit- 'falb' ~ harita- 'gelb, grün', from the word haray- 'gelb, falb', and the corresponding juxtaposition of the suffixes -it- and -ita- in Old Iranian, e. g. Avestan masit- 'ausgedehnt, groß' ~ masita- 'groß, umfangreich', from the word mas- 'lang, ausgedehnt, groß'. As to the suffix -ka-, this is one of the most frequent elements of word-formation in the Iranian languages, occurring very often also in combination with numerous other suffixes. It appears e. g. in Sogdian in the following groups of suffixes: -(’)n’k < *-(a)-na-ka-, -’nyk < *-anayaka-, -’yχk < *-aiχaka-, -’yk < *-ayaka-, -’yn’k < *ainaka-, etc. (see Benveniste, Essai de grammaire sogdienne, II, 95 foll.). The situation is precisely the same in the Saka language. Here, too, the suffix -ka- is very frequent ("sehr verbreitet und bis in die Spätzeit lebendig" — says Konow in his Khotansakische Grammatik, 67), and is used also in combination with several other suffixes; see e. g. -naa- < *-naka-, -laka-, -līka-, etc. (Konow, op. cit., 68, 70). Traces of the group of suffixes -taka-are found, too, in one or two Ossetian words. Thus, the word säftäg 'Klaue, Huf' which is connected with the Avestan word safa- 'Huf' (see Miller, Die Sprache der Osseten, 94), goes back undoubtedly to an Old Iranian form *safa-taka- and contains the compound suffix -taka- supposed to be present in the name Maldaγ < *Marya-taka-. Thus the derivation of the name  from the Old Iranian form *marya-taka- becomes very probable, and if the restoration of the name proves to be correct, it supplies another example of the phonetic change -ry- > -l-.

Old Iranian *ar-y- > ir-.

, Panticapaeum (193–208 A. D.). The name, certainly has to be read as irak- because, on inscriptions dating from the same period we often find η instead of ι: < Old Iranian *arya-ka- (Vasmer, Die Iranier in Südrußland, 39 foll., Iranisches aus Südrußland, 368).

Tanais,  No. 195 (188 A. D.). This name was regarded by Miller as a compound of the Ossetian words ir and vidag (ir 'Ironer (Ostosseten)' > *rya-; vidag, Munkácsi: KSz XXI (1932), 83: idagiedagä 'Wurzel', the approximate sense of the whole name being 'von arischer Wurzel'); Vasmer was right, however, in pointing out that the earlier meaning of the word vidag was probably 'Weide' (cp. e. g. Avestan vaētay- 'Weide, Weidengerte', and that with this meaning the name does not give any acceptable sense (Vasmer, Die Iranier in Südrußland, 41). With regard to , the first element in the name, Miller's interpretation may be accepted in any case. As to the second element  (stem: ), there are two

Additional Notes

To p. 80. The name  is uncertain because it is only restored as  by Latyschew (cf. CIRB No. 199).


possible explanations: 1. ~ Avestan vaēδah- 'Besitz'. 2. ~ Avestan vaēδay- 'Gestalt, Form'. In the first case the Old Iranian form of the name would be *rya-vaiδah-, meaning 'der den Besitz der Arier ergriffen hat', or 'der einem Arier gebührenden Besitz hat'; in the second case, we may assume the existence of an Old Iranian form *rya-vaiδay-, meaning 'of Aryan form'. Whichever interpretation we accept, the name  shows the development of the initial group of phonemes ary- into ir-.

Tanais,  No. 196 (beginning of third c. A. D.). This name, hitherto unexplained, may be traced back to an Old Iranian form *ryakāna-, i. e. the word rya- supplied with the well-known patronymic suffix -(a)kāna- > -(a)γān (see e. g. Armenian Boyekan < Middle Persian Bōi + akān Hübschmann, Armenische Grammatik. I, Leipzig 1895, 33, Middle Persian ’rdw’ng’n: ardavānaγān 'Ardawānian, deriving from Ardawān, etc.). Thus the meaning of the name irγān may have been 'deriving from Aryan' or 'deriving from Ir' ( = an ancestor bearing the name of īrrya-). The assumption of the latter meaning is obvious, on the basis of names like  < īrak < furīran, etc.

Tanais,  No. 195 (188 A. D.): the first element in the name goes back to Old Iranian *ryana- or *rya- (cp.  Vasmer, op. cit., 31).

Tanais,  No. 194 (225 A. D.): rya- (Vasmer, op. cit., 41, Iranisches aus Südrußland, 368).

Tanais,  No. 386 (225 and 212—229 A. D.) < Old Iranian *paru-ryana- 'viele Arier beherrschend' (Vasmer, Die Iranier in Südrußland, 55, RLV XII, 245).

Old Iranian *ar-y- > il-.

Olbia, Vasmer, Die Iranier in Süddrußland, 39. This name, hitherto unexplained, probably goes back, assuming the development *ar-y- > il-, to an Old Iranian form *aryaman- which may be compared with Avestan airyaman- 'Genösse' ~ Old Indian aryaman- 1. 'Genösse', 2. 'Name eines Āditya' (see on this point Bartholomae, AirWb. 198 foll., Benveniste: JA CCXXI (1932), 124 foll.) as well as with Middle Persian ērmān and Modern Persian īrmān 'Gast' (with regard to the latter see Horn, Grundriß der neupersischen Etymologie, Strassburg 1893, 32 foll. and Hübschmann, Persische Studien. Straßburg 1895, 20 foll.).

Names like , etc. indicate that we have to do here with an i-epenthesis; so the line of development is ir- < *air- < *ary-. The other two groups of names show, however, that this development was not general, but was restricted to a definite dialect or group of dialects. Thus one cannot derive, for instance, the element  in the name  from the Iranian form *airya-, as Vasmer did, (RLV XII, 244), because the regular development of this form is ir- or il-. This would be all the less justified as the i- and u-epenthesis can be shown to have existed, among the Old Iranian languages, only in the Avesta [81], and even here it is probably due only to the carelessness of Persian and Parthian scribes [82]. There can be no question of an i- or u- epenthesis common to all Iranian languages or going back to Proto-lranian. On the other hand, the testimony of these names indicates that

Additional Notes

To p. 81. Zgusta (Die Personennamen griechischer Städte der nördlichen Schwarzmeerküste. 232 foll.) would like to eliminate all instances of the phonemic change -ry- > -l- in Sarmatian. He says: "Es kann sich auch um verschiedene Aussprachen handeln, die in der Schrift den Ausdruck fanden." This is hardly possible. We can reckon with two possibilities: 1. The Sarmatian dialects had only the phoneme /r/. In this case no alternation between r and l was possible. 2. The phoneme /l/ arose as a result of the phonemic change -ry/ri- > -l-. In this case an alternation between r and l can perhaps be assumed, but it proves already the phonemic change -ry/ri-> -l-. Another instance for -ry- > -l- can be recognized in the name , which can be traced back to an Old Iranian form *čiryakāna-.

81. Bartolomae, Grd. d. i. Ph. I, 1 76.

82. Reichelt, Stand and Aufgaben der Sprachwissenschaft, 278. This view is represented by Bartholomae and his followers. A fundamentally different approach is seen in Andreas and his school, recently also in Bailey's theory (Zoroastrian Problems in the Ninth-Century Books, Oxford 1943. I 7 foll.). Neither theory does however, envisage epenthesis in Common or Prolo-Iranian. For a recent view on the whole subject sec Altheim, Literatur und Gesellschaft im ausgehenden Altertum, II, 189 foll.


epcnthesis must have been a fairly early phenomenon in some of the dialects and the existence of such dialects must, in fact, be assumed. Accordingly, one may actually suppose that the phenomenon did, in fact, exist in the language of the Avesta, in the first centuries A. D., as Reichelt (loc. cit.) assumes on other grounds. In any case, the dialect which supplies us with these names is connected, by means of this phenomenon, with those Northern and Eastern Iranian dialects in which the i-epenthesis can be shown to have existed: Saka ysīḍaa- < *zaritaka-, Afghan šil ~ Avestan vīsaiti, Ossetian innä < *anya-, Shughni. nir < *narya- [83].


83. With regard to these dialects see Reichelt, loc. cit. and Grdr. d. idg. Sprach- und Altertumskunde, II, 42, 33.

[Previous] [Next]
[Back to Index]