ܗܰܙܳܪܦܳܬ hzrpt hazārpāṯ χιλίαρχος chiliarch, lord of a thousand
OP *hazārapati- (Hinz 1975, 120 reconstructs OP *hazahrapati-, Median, on the basis of Gr. ἁζαραπατεῖς, but see below); MP [hzʾlpt] hazārbed chiliarch, MP Inscr. <hzʾlwpt>, Parth. Inscr. <hzrwpt> (KZ MP 29 and 31, Parth. 23 and 25, cf. Huyse 1999/2, 133 f.); Bactr. υαζαροχτο, υαζαροχττο, υαζοροχτο, scil. /hazāruxt/ (Davary 1982, s.v.); Arm. LW hazarapet and hazarawuxt (Hübschmann AG 174, no. 328; Garsoïan 1989, 531 f. with bibl.); Gr. ἁζαραπατεῖς (Hesych.: οἱ εἱσαγγελεῖς παρα; Πέρσαις) and ἁζαβαρίτης (Ctesias): but perhaps these Gr. loanwords are unconnected with OP *hazahrapati- and derive from OIr. *āara-pati-: see Szemérenyi 1975, 354–392 and Hinz 1999/2, 134 n. 222; ἁζαρίπτης, ἁζαροπτ (KZ Gr I 56 and 61, cf. Huyse 1999/2, 133 f.). See also the allotropes of the name of a Persian general Gr. Byz. Ἀζαρέφθας (Menander Protector, frg. 23, 11, 1 ed. Blockley 214 f.), Ἀζαρέθης (Procopius 1, 17, 1; etc.); Ἐξαράθ (Johannes Malalas 18, p. 461 l. 10 ed. Dindorf). Cf. Justi 1895, 88 and 128. Note that in Syr. there is also kylyrkʾ (LS 329b; allotropes: klyrkʾ, klyʾrkʾ), namely the transcription of Gr. χιλίαρχος (Aeschyl., Xenophon etc.), which alternates with χιλιάρχης (Herodotus); in its turn, the Gr. word is a structural calque on OP *hazārapati- ● am 2, 318, 5 ◆ LS 174a; PS Suppl. 99
The scripts MP <hzʾlwpt> and Parth. <hzrwpt> in KZ are generally intepreted as representing MIr. hazāruft. The traditional explanation (Szemerényi 1975, 357; cf. Khurshudian 1998, 80 f.; Huyse 1999/2, 133 f.) of the presence of the two allotropes MP hazārbed and hazāruft, both tracing back to OIr. *hazāra-pati-, is based on the hypothesis of two different stress patterns: *hazā́ra-páti- > MP hazārpat, > hazārbed; whereas *hazārá-pati- > hazārápt > hazāraft (with loss of the post-tonic vowel), then changed into hazāruft by metaphony induced by the neighbouring labial plosive. Consequently, Khurshudian 1998, 81 (and 283 n. 11) believes that Arm. hazarapet could have been borrowed from Parthian before the form was influenced by MP hazāruft, that is, before the third cent. AD; the less usual Arm. hazarawuxt would have been borrowed later from MP hazāruft.
This traditional explanation was criticized by Mancini (1988, 81 ff.), who denies that MP hazāruft has ever existed: the «w» between the two elements of the compound would be a merely graphic phenomenon, typical of the epigraphic script pārsīk and unattested in genuinely Parthian documents. This conclusion would be confirmed by the comparison with the Pahlavi Books, in whose script this «w» originally appears only in some compounds in -bar «-bl», but does not in the compounds in -pat «-pt». For these latter ones, the scholar supposes a usage typical of the MP inscriptions, that originated from cases in which the «w» was etymological, even if not yet pronounced. Hence, by way of analogy, the «w» would have been extended also to compounds in -pat «-pt», where it was etymologically unjustified.
According to Mancini (1988, 84 n. 32), the two scripts MP <hzʾlwpt> and Parth. <hzrwpt> of KZ represent MP hazārpat (> hazārbed), as well as the Syr. form hzrpt; therefore, the Gr. form αζαροπτ in KZ would not represent an old pronunciation, but it would be a pure transliteration of the corresponding Iranian forms. However, the two forms Bactr. /hazāruxt/ and Arm. hazarawuxt, in which -xt- is of uncertain origin, remain unexplained. Szemerényi (1975, 358) believes that MP hazāruft changed to hazāruxt in some not more clearly specified northern Iranian dialect, and was then borrowed into Armenian and Bactrian. In the same way, Mancini (cit.) holds as probable that an eastern allotrope hazāruxt was present in Parthian, whence it entered Arm. and Bactr. See also Benveniste 1966, pp. 67–71; Henning 1965a, 81. See also Schmitt, forthcoming.