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


3. The Sarmatian Dialects of the North Pontic Region

The Old Iranian diphthong au developed along two lines: 1. au (αο). 2 ō (ω).

Old Iranian *au > au.

'Sarmatian tribe'. This name has been interpreted in different ways. Jacobsohn thought that the word ryana- was somehow concealed behind it; later, however, he gave up this conjecture (Arier und Ugroffinen, 234, 257). Miller ( 1886 October 235 — article inaccessible to me) and Tomaschek (SWAW CXVII [1888], 37, PWRE I, 2660) connected the name  eith Avestan auruša- 'weiß', Ossetian ors, ūrs 'weiß'; this interpretation was later accepted by Vasmer (Die Iranier in Südrußland, 32) and Altheim, too (WaG II [1936], 319).

Against this interpretation Marquart has tried to explain the name in a new way. The starting point of his new interpretation was the existence of historical contacts between Alans and Aorsi. According to the testimony of Chinese sources, Yen-tsai whom Marquart, following Gutschmid and Hirth, identifies with the , changed his name to A-lan. Since, however, Greek and Latin sources inform us that in Eastern Europe the name Aorsi was replaced by Alan, Marquart comes to the conclusion that the name  is but the earlier name of the Alans. He now attempts to establish the meaning of the name Alan, calling to aid the series of epithets applied to an Armenian nobleman's family in Faustus Byz. (4, 2): ałanazgik‘, alana-drawškarcowēnšankwaržnakanišk‘. The last of these four epithets, in Marquart's view, goes back to an adjective *waržnak which may be a borrowing of a Middle Persian form *waržānak or *waržēnak (derived from Middle Persian warž, Modern Persian war 'Größe, Würde'): hence the meaning of the word is 'würdig'. The expression arcowēnšank‘ is purely Armenian, with the meaning 'Adlerstandarten führend'. The second element in the first epithet is the Armenian word azg 'Geschlecht, Nation', hence the compound probably means 'aus ałanischem Geschlecht stammend'. Finally the second element in the second epithet is the Armenian word drawš (< Iranian drafš) 'Banner', so that the meaning of the compound is 'ałanische Banner führend'.

Since thus all the epithets express worth and dignity and are closely related in meaning. Marquart was justified in concluding that the word ałan, i. e. the initial element in the first two epithets must mean 'siegreich, ruhmvoll, würdig'. Hence, according to him, "der Volksname Alanen wird demnach ein Ehrenname sein, den sich das Volk selbst beilegte und der eine Gruppe verschiedennamiger iranischer Nomadenstämme der kaspisch-pontischen Steppen zu einer politischen Einheit zusammenfaßte".

Regarding the word  as the former name of the Alans, Marquart then

Additional Notes

To p. 82. Zgusta (Die Personennamen griechischer Städte der nördlichen Schwarzmeerküste. 263 foll.) would like also to cancel the name , but his argumentation is only based on a series of misunderstandings. He calls in doubt the possibility of a transcription αο of the diphthong au. But this was the only possible correct way of transcription because αυ in Greek was already monophthongized. Then he quotes the form Avorsorum from Tacite but such a reading does not exist (cf. also F. Altheim: Geschichte der Hunnen. I. 70, 74 foll, on the different readings of the name in the codices of Tacite). Zgusta questions even the existence of the Ossetian word urs 'white' and he asserts that the Osseto-Russian dictionary by Kasaev does not contain it. On p. 344 foll, of Kasaev's dictionary, however, the word urs together with his numerous compounds can be found. Otherwise I already discussed all these problems in the 1st edition in a detailed manner.


proceeds to look for a similar meaning behind it. For the purposes of interpretation he distinguishes, first of all, two forms of the name: 1. Arsoae (Tab. Peut. IX 5, X 1), *Arzoae (Abzoae: Pliny, Nat. Hist. 6, 38) — 2.  (Strabo, Ptolemy), Aorsi (Pliny Nat. Hist. 4, 80). Of these, Marquart derives the form Arsoae, Arzoae from the Iranian form *arž-awa- (cf. Modern Persian ar 'Wert', Avestan arəah-, Middle Persian ar, etc.); as to the form Aorsi, he interprets it as the Iranian compound *hu-arž- (with the approximate meaning 'guten Wert habend'?). Thus Marquart concludes that "der Name Aorser, ebenso wie die Alanen, eine ehrenvolle Selbstbezeichnung ist, welche sich das Volk bezw. der führende Stamm wahrscheinlich bei der Begründung einer größeren politischen Einheit beilegte" (Untersuchungen zur Geschichte von Eran, II, 82—86.)

Marquart's explanation is, both historically and linguistically, so well-grounded that one cannot simply pass it by. The first question in this connection is whether we really have to do with two forms of the name. In support of his conjecture. Marquart refers to the parallel forms Su-gambri ~ Gambrivii and Wisi-gothae ~ Guto-nes. But these parallels only demonstrate, in a general way, the possibility of a tribe or a people having two names, differing only in an element implying comparison. The question whether this applies also to the name  can be decided only after a careful investigation of the data containing the name.

An examination of the ancient sources referring to the name  yields the following result: of the two forms, separated by Marquart, it is only the , Aorsi that can be substantiated by sound textual tradition. Textual tradition supports unanimously the form  in Strabo and Ptolemy, and nearly unanimously the form Aorsi in Pliny (Nat. Hist. 4, 80). In the case of Tacitus (Ann. XII 15. 16 and 19), the MSS give the forms adorsorum, adorsi, and aduorsorum. These forms are obviously due to the circumstance that the scribes wanted to impart some meaning to a name unintelligible to them. But these particular forms of 'rationalization' may all be traced back to the form Aorsi, not to Arsoae. Thus the data in Tacitus, too, are organically connected with the form , Aorsi.

As to the other form distinguished by Marquart, we have to point out, first of all, that the Plinian form Arzoae is only Tomaschek's conjecture: the MSS give Abzoae which is probably a corruption — yet there is no material proof whatever to show that the name Aorsi is hiding behind it. Hence this conjecture may be left out of account. But if this is so, the forms Arsoae in the Tabula Peut., by themselves, do not possess any special significance. The Tabula is full of corrupt forms, so that it would be contrary to all rules of scientific methods if one were to regard the form Arsoae as authentic as the forms , Aorsi found in Strabo, Ptolemy, Pliny, and Tacitus. Moreover, judging from the character of textual corruption due to the copying of MSS, one may demonstrate almost palpably the corrupt nature of the form Arsoae as well as the causes of its origin. The deviation in the endings of the two forms Arsoae and Aorsi may be easily explained by the supposition that, parallel with the Latinized form Aorsi of the name , the geographical literature of the Romans also used the form Aorsoe, a transcription of the Greek name.

There are plenty of examples in Roman authors for such parallel usage in names taken over from Greek geographical literature. Thus, e. g., the name  appears in Mela (II, 1) as Neuri, while in Pliny we find the form Neuroe (Nat. Hist. 4, 88); conversely, the name  is transcribed by Mela (II, 1) as Hamaxobioe, by Pliny (Nat. Hist. 4, 80) as Hamaxobii; similarly, the name  occurs in Mela (II, 1) as Arimaspoe, while in Pliny (Nat. Hist, 4, 88) we find Arimaspi. It may be observed that such un-Latinized names, transcribed from Greek, often have their


ending -oe distorted, or rather Latinized, to -oae in the course of MS transmission. Thus, e. g., in some MSS of Pliny the name Enoecadioe (Nat. Hist. 4, 83) appears as enocadloae, enoae.adioae. In the latter form both Greek diphthongs -oe- have been 'corrected' to -oae- by the copyist. Similarly, some MSS of Pliny give the form neuroae for the name Neuroe (Nat. Hist. 4, 88). Thus the ending of the form Arsoae in the Tabula Peut, may be easily explained as the result of a secondary Latinization of the name Aorsoe. As to the deviation in the initial sounds of the word (Aor- ~ Ar-), the omission of one of two juxtaposed vowels is a frequent phenomenon in the transmission of MSS. Thus the name of the people given by Mela as Choamani is found in several of Pliny's MSS as comani (Nat. Hist. 6, 48). Another example from the transmission of Pliny's text is the distortion of the word Bactros (Nat. Hist. 6, 47), first, to baotros, then its further corruption to botros. On the basis of these examples we are justified in taking it practically for granted that the form found in the Tabula Peut, is a distorted form of the Latin transcription of the name . The process of its origin may be outlined as follows: *Aorsoe > *Aorsoae > Arsoae.

Thus the thesis which forms the base of Marquart's edifice of explanations — viz. the existence of two forms for the name of the Aorsi — has proved to be unacceptable. The other fundamental question which has to be posed in connection with Marquart's theory, is whether the name  may, in fact, represent the transcription of an Iranian form *hu-arž-. Since the Greek letter o stood for a definitely close o-sound (= ) it is most probable that the name  represented a foreign form *aurs. Such a form is, of course, very far from Marquart's *hu-arž-, the Greek transcription of which would be  or . Marquart himself was aware of the grave difficulties which arise in this connection: hence he gave several parallels to illustrate the possibility of transcribing as  the conjectural form *hu-arž-. His examples are as follows:

< Iranian *hu-warna-, 'wohlbewehrt', from the stem war- 'wehren'.
= Avestan hutaosā- 'EN der Schwester und Gemahlin Vištāspas’.
, Lycian Humrkkā = Old Persian haumavarga-.

Finally, Marquart quotes several names beginning with , in which the first element represents Old Persian *wāta-; e. g. — Old Persian wātafraδāta-.

But these examples are either not suitable parallels to the transcription of Iranian *hu-arž- as Greek  or have to be interpreted in a way different from Marquart's.

The name  is certainly not the transcription of an Old Iranian form *huvarna-: it may either stand for *əurn, a conjectural development of this form; or it may be connected (as Tomaschek suggested in PW-RE I, 2659) with a quite different word, viz. Old Iranian āvarana- 'Schutzwehr', the existence of which may be conjuctured on the basis of Old Indian āvaraṇa- 'verhüllend; Verhüllung, Hülle, Decke, Gewand': in this case the Greek form would transcribe the development *āurn.

The name  may also represent the transcription of *ətōs, a later development from Old Iranian *hutausā-.

Greek  and Lycian Humrkkā reflect different developments of Old Iranian *haumavarga-: 1.  < *əmurgi, *əmurg < *hauma-vargah; 2. Humrkkā < *hūmūrga < *haumavargah.

From the angle of phonemics, the correspondence ~ Vātafraδāta implies quite a different problem from that represented by  ~ *hu-arž-.


This is a case of the Greeks replacing a group of sounds (va-, a-) absent from the phonemic system of their language, by another phoneme or group of phonemes (αυ). Similar cases are very frequent in the Greek transcription of Iranian names (see e. g. Harmatta, Ant. Hung. II, 35).

We may thus establish that the Greek form of the name  cannot represent the transcription of an Iranian form *hu-arž-. The latter may have developed into *hvarž, *χvarz, *varz, *χarz, *χaz: but all these are far from the foreign form *aurs, the existence of which may be conjectured on the basis of the spelling  Thus the form *aurs hiding behind the name can hardly be anything else but an intermediate stage in the following development: Old Iranian *aruša- > *auruša- > Ossetian orsrs. We may also remark that the name  shows u-epenthesis, a phenomenon which connects it with names like , etc. showing i-epenthesis.

Tanais,  No. 79 (225 A. D.). According to Vasmer, the word may be an abbreviated form of a person's name like Avestan gaodāyah- 'das Rind hegend und pflegend' or gavayan- 'der Rinder hat' (see op. cit. 36, RLV XII, 244) Since the group of phonemes -ava- is usually transcribed in the names of the inscriptions as -av- or as -αvα-, -αvo- (e. g. ~ Avestan yava- 'Getreide', Vasmer, Die Iranier in Südrußland, 55; G.  ~ Avestan syāva- 'schwarz', Vasmer, op. cit., 51, etc.), probably only the first possibility has to be taken into account.

Old Iranian *auō.

Panticapaeum, Latyshev, IOSPE II, 107 ~ Avestan raoχšna- 'licht, glänzend' (Vasmer, op. cit., 49).

Tanais,  No. 85 (220 A. D.). According to Vasmer (op. cit., 37) this word has been obtained by suffixation from the short form of a name derived from *gōš, the equivalent of the Avestan word gaoša- 'Ohr'. Besides Vasmer's conjecture there is also the possibility that this name has simply to be regarded as an equivalent of Old Persian *gaušaka- 'Horcher', Parthian *gōšak (> Armenian gušak 'Angeber, Denuntiant'). [84]

Panticapaeum, Latyshev, IOSPE II, 29: the word represents the short form of a name like Avestan aspāyaoδa- (zu Rosse kämpfend' (Vasmer, op. cit., 41). A similar name has recently been found in Sogdian: ywδrzmk: yōδrazmag- (see Reichelt, Die soghdischen Handschriftenreste des Britischen Museums, II, 56).

84. With regard to these see Schaeder, Iranica, 5.

[Previous] [Next]
[Back to Index]