We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Published online by Cambridge University Press: 24 December 2009
The corpus of Bactrian, of which KI forms the most important part, is too restricted to enable us to form even a remotely complete picture of its phonemic system and development. This would be the case even if every word had been finally or at any rate plausibly interpreted. As long as we restrict ourselves to the SK inscriptions, the language seems to represent, in the main, one single dialect. But if we include in the term Bactrian everything written in Ir. language and Greek script, from the coins of Kadphises I down to the IT of A.D. 866, it seems reasonable to assume that dialect variants, of chronological or local nature, are to be found. In this paper I shall be dealing mainly with the KI, with only occasional references to other sources. For the interpretation I shall base myself on the work of Henning, Benveniste, and Gershevitch, together with the contributions made by Harmatta and Humbach; and it will be assumed that the reader is conversant with, above all, Henning's interpretation of a substantial part of the text of the inscription.
1 I shall be using Humbach's abbreviations, Baktr. Sprachdenkmäler, i, Wiesbaden, 1966Google Scholar: IT = Tochi inscriptions; KI = Kaniska inscription from Surkh Kotal; Kpl = (so-called) Kaipur inscription from SK; MB = Berlin MS fragments; NumK = Kusāna coins; Pal. I = Palamedes inscription from SK; Par. I = wall inscription from SK; Sig. = seals.
2 Note, however, Ir. xš- in Šao and xšono, etc.
3 ‘The Bactrian inscription’, BSOAS, XXIII, 1, 1960, 47–55Google Scholar (references given to page); his other articles will be quoted in full.
4 ‘Inscriptions de Bactriane’, JA, CCXLIX, 2, 1961, 113–52Google Scholar (references given to page).
6 Review, BSOAS, XXVI, 1, 1963, 193–6Google Scholar; ‘The well of Baghlan’, Asia Major, NS, XII, 1, 966, 90–109Google Scholar; ‘Bactrian inscriptions and manuscripts’, IF, LXXII, 1–2, 1967, 27–57Google Scholar.
6 The great Bactrian inscription, Budapest, Akadémiai Kiadó, 1964Google Scholar (references given to page).
7 If Humbach, 100, is right in reading the end of Kpl xšono dbo uu(bdo) (cf. his Teil II, Tafel 21), corresponding to the beginning xšono soe/θ ‘year 275/9’, we have dbo ‘two’, which would agree, not with Mj.-Yd. lu, lo(h), but rather with Psht. dwa (Sherranī dba). For a tentative, but quite hypothetical, explanation of this dw- as contrasted with bəl ‘second’, war ‘door’, and las ‘ten’, cf. EVP, s.v. dwa. There is no reason to believe that Greek l represents an attempt to render Ir. δ.
8 Or, possibly, /č/. Many, but not all EIr. languages have c > č, and Greek sigma would perhaps lend itself more readily to rendering a dental affricate.
9 cf. Bailey, , BSOAS, XII, 2, 1948, 329Google Scholar, and Henning, , BSOAS, XVIII, 2, 1956, p. 367Google Scholar, n. 2, regarding Icsuna < Olr. *xšaivana-.
10 Gershevitch, , IF, LXXII, 1–2, 1967, 47Google Scholar, suggests nokonziki = nigundzig, cf. Skt. nikuñc-.
11 Shgh. wixkamb- ‘to tease wool’ < *skumb- is semantically too remote.
12 JA, CCXLVI, 4, 1958, 363Google Scholar.
13 Humbach's readings (n)ibixto, and nibigo ‘Schrift’, from the IT are much too fanciful for any conclusions to be drawn from them.
14 Is it certain that Henning, , BSOAS, XVIII, 2, 1956, 367Google Scholar, is right in deriving Baγlān < *Baγlān(g) < *baga-dānaka-? Could there not have existed a side-form *-dāna- without the -ka- suffix ? I know no other example of *-ān(g) > -ān.
15 If, with Harmatta, 457, < *barg- ‘to wish’.
16 For froxorto, cf. below.
17 If xirgo(mano), with Harmatta, 467, should go back to *x(a)rika- ‘stone’, we may compare Yazg. xərik ‘sand’, Sar. xarayj ‘coarse sand, gravel’ (but note Sar. x-, not š-!). A derivation from xar- would not entail any metathesis. But Benveniste and Gershevitch are probably right in taking the word to be a proper name.
18 cannot agree with Gershevitch, , AM, NS, XII, 1, 1966, 102Google Scholar, in deriving porogato < *parikata- (with t representing -d-), and in taking, AM, 106, porooato as a 3rd pers. sg. subj. in *-ād.
19 cf. also Prs. pargār ‘circle’.
20 IIFL, II, 81, 155.
21 Shgh. has zῙrd, with contraction after umlaut and -t- > -d-; Wkh. zart tells us nothing; Sgl. zāl treats *-rit- in the same way as *-rt-, and Ishk. zord is either, like Shgh., with *-r-d-, or, more probably, a very early loan from Prs. (in other loan-words Prs. a is retained).
22 v. IIFL, II, 500 f.
23 cf. mât ‘killed’, höt ‘heard’ < *mārita-, *harwita-, rather than with IIFL, i, 38 < *marxta-, *harwta-. Cf. also Par. ânt ‘brought’ < *ā-nǐta-, γunt ‘found’ < *windita- (γun-).
24 Pakhalina gives the variant forms důld, with secondary -d, and důrt, formed from the present base, like Sköld's dēr-d.
25 v. below.
26 But Bailey, , KT, vi. Prolexis to the book of Zambasta, 140Google Scholar, gives nähvarr- ‘to grasp at, desire’, buhurs- ‘to grasp, appropriate’.
27 Regarding nisto, v. below.
28 P. 54, n. 10, he writes: ‘One might make bold to emend froxoaš- [of all three versions, G.M.] into froxoar-’, and suggests a possible connexion with Khwar. xwarδ- ‘to flee’as a derivative of xwar-. I am afraid this would not help us much.
29 v. above, ad xoto.
30 JA, CCXLVI, 4, 1958, 361Google Scholar.
31 Gershevitch, GMS, §343.
32 Benveniste, , Langue ossète, 46Google Scholar.
33 This must be a misprint for marto given a few lines above, just as also mašto (< *maxšta-) must stand for maxto.
34 Vend., xviii, 28, MSS have also -m 
lžda-.
35 The alternative suggested, p. 55, n. 7, of a derivation from masita- ‘great’(adopted by Humbach) is excluded for phonetic reasons.
36 But Orm. K məz- ‘to be twisted’, muz-aīm ‘I twist, spin’ < *mazj- < IE *mezǵ-.
37 cf. Pal. I, end (Humbach, 102) (n)obixto mo mašto.
38 Gershevitch, , IF, LXXII, 1–2, 1967, p. 32, 1. 8Google Scholar, proposes an alternative derivation < *ham-āsrita- ‘ladder’. This is a very ingenious and tempting explanation; but we know too little about the development of sr and of the eventual syncope of i to be able to accept it as the final solution of the problem, in preference to the simpler one < *mašta-.
39 cf. the examples given in IIFL, n, 49, 73.
40 cf. Gershevitch, , AM, NS, XII, 1, 1966, 103Google Scholar, m'andar, or māndar.
