Re: [tied] Thracian place-names

George Hinge wrote:

> In Classical Greek, Dnepr was called, as we all know Borysthenes, but
> lateron the form Danapris came into use (Pseudo-Arrian, Periplus
> §58 Borusqe/nhn potamo\n nausi/poron, to\n nu=n Da/naprin lego/menon;
> cf. also §§ 60, 91). This points to a Scythian (or Sarmatian or
> Alanic) *Dan-apra-.
>
> Abaev explains the second part of the word as a derivation from
> Iranian *a:p- "water", and he will find the same derivation in
> Ossetic arf "deep, depth" (Istoriko-etimologic^eskij slovar'
> osetinskogo jazyka, vol. 1, 63). The same word may be the first part
> of the Scythian mythical name Arpoxais, i.e. "water-king" (the
> ancestor of the two common tribes, cf. A. Ivantchik, Une légende sur
> l'origine des Scythes, REG 112 (1999) 141-192, G. Hinge, Scythian and
> Spartan Analogies, in: The Cauldron of Ariantas, Aarhus 2003, 55-74).

The comparison of <arf> with "Arpoxais" is fanciful. It would require
the Ossetic metathesis of *-fr- > *-rf- (*afra- > arf) to date back,
impossibly, to Old Iranian times or to have happened in Scythian
independently. But if so, why should we have an unmetathesised -pr- in
the hydronym? By the way, Iranian *a: was long in the nom.sg. of *a:fs^
(with the regular nominative lengthening) and in the "strong" cases
(presumably because Brugmann's Law applied there: *h2op-es > *a:pah,
etc.), but there's no reason why it should have been long in *afra- <
*h2ap-ro-.

> Is it possible for Iranian *a: to turn up as Slavonic *e:? We have
> the exact same problem with Dnestr, which is written Danastris in the
> late Greek texts (= classical Tyras).

Abaev offers no explanation of the *e^, and that's a major weakness of
his etymology. Another possibility perhaps worth contemplating is
dialectal Iranian "umlaut" applying across a consonant cluster, e.g.
*-afri- > "Sarmatian" *-aifr- --> Slavic *-e^prU. The i-stem *-afri- as
the compositional variant of *-afra- seems OK to me.

Piotr