Introduction
The Bactrian language attests two words for ‘year’. While the spectacular discovery of new manuscripts in recent decades have now attested Bact. σαρδο, σαρλο (cf. MBact. srdʾnyg ‘pertaining to years’) continuing the common Proto-Iranic word for ‘year’ *sardV- (cf. Av. sarǝd-, OPers. θard-, Chor. srδ, Sogd. srd, etc.),Footnote 1 the term χϸονο was previously attested in dating formulae in the Surkh Kotal inscription, a handful of other brief inscriptions and coin legends. While this was previously considered the Bactrian word for ‘year’,Footnote 2 we now know that the term χϸονο, rather than being the general term in Bactrian, designates a more circumscribed concept of ‘(calendar) year, (regnal) year’ in the vast majority of texts pertaining to the year count of the Bactrian era.Footnote 3 The etymology of Bact. χϸονο remains uncertain and the only suggestion considered worth noting by Nicholas Sims-Williams in BD2 repeats the suggestion of Andreas Thierfelder recorded by Helmut Humbach in Baktrische Sprachdenkmäler, who considered the possibility of a loanword from Hellenistic Greek χρόνος ‘time’. In this article I would like to briefly revisit this etymological suggestion and ultimately argue that it should be rejected on grounds of formal comparison. It gives me pleasure to offer this brief study to François de Blois, with whom I studied Bactrian and other Middle Iranic languages at the Ancient India and Iran Trust, Cambridge. Given François's interests in Bactrian chronology and the linguistic interactions between the Iranic languages and their neighbours, I hope that he will find this short article a fitting contribution in view of his work and philological erudition in these areas.
Early etymological proposals
As observed by Humbach, the initial discovery of Bact. χϸονο ‘(calendar) year, (regnal) year’ added another word to the group of vocabulary designating periods of time alongside Khot. kṣuṇa- ‘section, period of time; period of rule in a year’ and Prak. kṣuṇa- ‘time’.Footnote 4 It remains a reasonable assumption that the Bactrian, Khotanese, and Prakrit lexemes are connected to each other in some way because of their close similarity in form and semantics. In the initial publication of the Surkh Kotal inscription André Maricq connected the lexeme to the Khotanese form and referred to the etymological proposal of Sir Harold W. Bailey who reconstructed for it OIr. *xšaiwana- ‘ruling, reign’ derived to the verbal root *xšai- ‘to rule’ (cf. Ved. kṣay- ‘to rule, have power over’).Footnote 5 With the evidence available at the time of Baktrische Sprachdenkmäler I (1966), Humbach ruled out Bailey's etymology on the grounds that (if one assumes that Khot. kṣuṇa- and Bact. χϸονο are true cognates) other well-known examples of PIr. initial *xš- appear to be reflected normally in Bactrian as ϸ- ~ /š-/, citing the well-known examples ϸαο, ϸαυο ‘king’ (< PIr. nom. *xšāwā to n-stem *xšāwan-) and ϸαρο, ϸαυρο ‘city’ (< PIr. *xšaθra-) and their derivatives.Footnote 6 As an alternative Humbach accepted the suggestion of his Mainz colleague Andreas Thierfelder that Bact. χϸονο could be a loanword from AGk. χρόνος ‘time’ with a specialised meaning ‘year’ developed within Bactrian from which the Khotanese lexeme would have been a loanword from Bactrian.Footnote 7
In light of the new Bactrian documents from northern Afghanistan we may well reconsider Humbach's generalisation that PIr. initial *xš- regularly yielded initial ϸ- in Bactrian. In favour of this interpretation we may now add the following examples:
1. ϸιι-, ϸι-, ϸ- ‘to be able, can; must, ought’ < *xšāya- (BD2 pp. 284-85, EDIV pp. 451–52)
2. ϸιζγο ‘good; well (in health)’ < *xšiǰa-ka- (BD2 p. 284, cf. EDIV p. 456)
Two further examples of initial xš- may also be adduced, but their reflexes are likely conditioned by intervening consonants in their complex onset clusters.Footnote 8
3. αχνωρο ‘satisfaction, gratitude’ < *xšnauθra-, cf. Av. xšnaoθra-, MMP/Parth. ʿšnwhr (BD2 p. 199)
4. χοατο ‘sixty’ < *xšwašti-, cf. Av. xšuuašti-, Chor. ʾxyc (BD2 p. 279)
A small number of possible counterexamples to the proposed general development of initial PIr. *xšV- > Bactrian ϸ- also exist:
5. χαβρωσο ‘(by) night and (by) day’ < *xšapā- ‘night’ (cf. Av. xšāp-, Chor. (ʾ)xyb, Sogdian xšp-) + ρωσο ‘day’ (BD2 p. 276)
6. χαρο (title) ‘khar, ruler’ < *xšāθriya- (cf. BD2 p. 277)
Perhaps also the following, but without any clear Old Iranic antecedents:
7. αϸχαλο, αϸαχαλο noun ‘grace, indulgence’ if from *xšadV-, cf. Parth. ʾxšd ‘mercy’ (BD2 p. 200)
The first two counterexamples could possibly be explained away as later dialectal loanwords,Footnote 9 but in any case there a good parallel does not appear to exist to support a development of initial PIr.*xšV- > Bact. χϸV- and Humbach is likely to have been correct to suggest that the origins of χϸονο should be sought elsewhere. To my knowledge, no alternative etymology for Bactrian χϸονο apart from Thierfelder apud Humbach has yet been suggested, nor has this hypothesis of a borrowing from AGk. χρόνος been adequately re-examined.Footnote 10 While I do not have any new suggestions regarding the former matter, in the remainder of this article I will undertake to investigate the latter.
Re-evaluating the Hellenistic Greek loanword hypothesis
Approximating the phonetics of Hellenistic Greek χρόνος
The reconstruction of the phonetics for a Hellenistic Greek source form likely to underlie χρόνος poses relatively few problems of interpretation as most of the phonemes in this word have remained relatively stable in Greek to the present day.Footnote 11 The only major obstacle is the thorny question of when the shift of the voiceless-aspirated stops /pʰ/, /tʰ/, and /kʰ/ to voiceless fricatives /f/, /θ/, and /x/ had occurred, and whether one should reconstruct [kʰró̞no̞s] or [xró̞no̞s] as the potential source form for the proposed Hellenistic Greek loanword. In the koiné of the Egyptian papyri there is no good evidence for early fricativisation of Ancient Greek /kʰ/ until after the Roman Imperial period.Footnote 12 Elsewhere Eduard Schweizer accepted evidence for fricative pronunciation of <χ> in the inscriptions of Pergamon already by the second century bce,Footnote 13 and Leslie Threatte considers evidence for <φ> representing /f/ in less literate Attic inscriptions of the second century ce.Footnote 14 On the basis of this evidence, while admitting that there is not much to go on, Geoffrey Horrocks has plausibly suggested that the fricativisation of Hellenistic Greek /pʰ/, /tʰ/, and /kʰ/ began in varieties of koiné spoken outside of Egypt and was carried to completion probably by the fourth century ce.Footnote 15 Certainly an upper date of the fourth century ce is too late as a terminus post quem where we can be reasonably confident of a form [xró̞no̞s] as a potential source form for a borrowing, since if χρόνος was the source of Bactrian χϸονο it must have been borrowed at least by the beginning of the Bactrian era, which has been calculated as 223 ce by Nicholas Sims-Williams and François de Blois.Footnote 16 And, even in consideration of that date, it must have been borrowed at some time before then given the earlier attestations in the Surkh Kotal and Rabatak inscriptions.Footnote 17 At the same time, however, we cannot entirely rule out that such a pronunciation was not in use in some varieties of koiné in the later Hellenistic and early Roman Imperial periods. Because of the uncertainty over whether this sound change would have occurred or not in the variety of Hellenistic Greek which could have been the potential source form for Bact. χϸονο, I consequently will consider both [kʰró̞no̞s] and [xró̞no̞s] further below as potential source forms in evaluating the possibility of a Hellenistic Greek loanword.
Approximating the phonetics of Bactrian χϸονο
The earliest attested form of the lexeme is χϸονο, and is first found with in a datable context in the Surkh Kotal inscription (Kushan era 31).Footnote 18 This form is regular in the earliest dated documents until the first attestation of αχϸονο in document J (Bactrian era 295 = 517 ce) and with the exception of document N (Bactrian era 407 = 629 ce) αχϸονο, this continues to be the normal form encountered in the later dated documents.Footnote 19 It appears clear, therefore, that the prothetic α- in attested in later Bactrian documents is a secondary phonological development. While the phonological representation of a Bactrian lexeme remains a matter of interpretation, we may reasonably postulate a pronunciation xš- for initial χϸ- and -n- for medial -ν-.Footnote 20 While we know more about the vowel system of Bactrian better than other Middle Iranic languages, the phonetic interpretation of the spellings of medial -ο- and final -ο are not entirely unproblematic. Georg Morgenstierne observed that in Bactrian epigraphic texts that /a/ is usually written before a nasal and concluded that it is probable that the medial vowel is -u-.Footnote 21 This observation fits well with the comparison of Khot. kṣuṇa-. The phonological status of final -ο is difficult to determine. Shortly after the publication of the Surkh Kotal inscription, Walter Bruno Henning suggested of the ubiquity of final -ο in the Surkh Kotal inscription that “it is possible omicron expressed a vowel actually pronounced in speech at the time of the inscription; in most cases it functioned virtually as a word divider”.Footnote 22 Alternatively, Morgenstierne considered it would be surprising that final vowels would have been completely lost at such an early period in East Iranic and considered it more probable that final -ο at least represented a reduced vowel [ǝ].Footnote 23 By the time of the late Bactrian documents at least, if there was a final vowel represented by -ο it appears to have been completely lost to judge from the orthography of the Manichaean Bactrian fragment.Footnote 24 In any case, I will therefore assume a probable phonetic interpretation of the earliest form of Bact. χϸονο as [xšunǝ] or [xšun].
Final Obstruents in Greek Loanwords in Bactrian
Only four certain loanwords of Greek origin are considered in BD2: Bact. διναρο ‘dinar’ < AGk. δηνάριον (ultimately < Lat. denarius) (BD2 p. 209); δραχμο ‘dirham’ < AGk. δραχμή ‘drachma’ (BD2 p. 209); Bact. οιϸοηγγο, οιϸοιγγο, οιϸιγγο ‘cloth made of linen or cotton’ < Bact. *οιϸο ‘linen, cotton’ < AGk. βύσσος ‘flax, linen’ (+ Bact. -ηγγο ‘suffix’) (BD2 p. 248); and Bact. σιμινο ‘made of silver’ (adj.), ‘silverware’ (n.) < Bact. *σιμο ‘silver’ < AGk. ἄσημος ‘unmarked, uncoined (silver)’ (+ suffix *-aina-) (BD2 p. 264). From these examples it would seem reasonable to assume that Greek nouns were normally borrowed into Bactrian as the uninflected stem, and a lack of final -ς in Bact. χϸονο is not a problem for the Hellenistic loanword hypothesis.
PIr. *xr- in Bactrian
To better understand how one might expect Bactrian to have adapted a foreign initial consonant cluster /kʰr-/ or /xr-/, I will now consider the regular reflexes of PIr. *xr- in Bactrian. As far as the available Bactrian documents attest, initial κρ- does not appear to have been a permitted onset cluster in Bactrian, but the lack of attestation of an initial cluster κρ- is probably not unexpected historically because PIIr. *k- regularly developed to *x- before a consonant in Proto-Iranic.Footnote 25 Words containing reflexes of *kr̥- from an original zero-grade appear to have been normally vocalised with an anaptyctic vowel, e.g. κιρδο < *kr̥ta- (BD2 p. 223, EDIV pp. 236–38), κιϸαγο < *kr̥šāka- (*kr̥ša- ‘to plough’, cf. Ved. kr̥ṣa-, Av. pairi.karša-; BD2 p. 224) etc. It is unclear to me whether the allophonic rule operating in Proto-Iranic which originally shifted voiceless plosives to fricatives was operating at the time when an alleged borrowing of χρόνος would have taken place, although it is notable that the only attestation of -κρ- as a consonant cluster in the Bactrian corpus occur in the names of Bodhisattvas in the Buddhist fragment za, βικραδο (BD2 p. 203) and σιγγοβικριδο (< Siṁha-vikrīḍita-, BD2 p. 263), which may be regarded as representations of Buddhist Sanskrit proper names and likely not subject to a Bactrian phonological rule. If the potential Greek source form for χρόνος was /kʰr-/, then we therefore might expect that it would be adapted into Bactrian with an initial /xr-/. If the potential Greek source form had already undergone the inner-Greek shift of /kʰ/ > /x/, then we should reasonably expect that it would have been adapted into Bactrian as /xr-/, or another similar permissible syllable onset.
I will now consider certain examples of reflexes of PIr. *xr in Bactrian gathered from the glossary of BD2, excluding obvious secondary derivatives:
• PIr. *xra- > Bact. αχρ-
■ αχριιανο ‘purchasable’(?) of unclear stem formation but clearly related to χιρ- ‘to buy’, αχρινο ‘purchase’ (BD2 p. 199)
■ αχρινο ‘purchase’ < *xray-anā̆-, to the root of χιρ- (BD2 p. 199)
• PIr. *xrī- > Bact. χιρ-
■ χιρ- / χιρδο ‘to buy, acquire, purchase’ < *xrī̆nā-/xrīta-, (cf. Sogd. xryn/xryt, Khot. ggän-; BD2 p. 227, EDIV pp. 446–47)
■ χιρηγο ‘purchase, purchase price’ < *xraya-ka-, cf. Ved. krayá- ‘buying, purchase’ (BD2 p. 227)Footnote 26
■ χιρσο ‘purchase, purchased (property)’ < *xrīti-čī-, cf. Sogd. xryc ‘purchase’ (BD2 p. 227–228)
• Medial PIr. *-xr- > -χρ-
■ αβαχρηγο ‘fee, compensation, wages’ < *apa-xraya-ka-, cf. Sogd. prxyy ‘wages’, NPers. barxai ‘compensation, ransom’ < *apa-xraya- (cf. Ved. krayá- ‘buying, purchase’) (BD2 p. 199).
■ οιχρηγανο, οιχαρηγανο ‘hire, rent’ < *wi-xraya- (Ved. vi-krayá- ‘sale’) + suffix -γανο (BD2 p. 248)
■ οιχρινο ‘hire, rent’ < *wi-xraya-anā̆- (BD2 p. 248)
Perhaps to be added to these is the verb of uncertain meaning φριχηϸ- ‘to molest’(?), ‘to seduce’(?) if from *fra-xrāšaya-, cf. NPers. xarēšīdan ‘to scratch’, Sogd. xryš ‘to irritate’ < *xrāšaya-, Chor. bxrʾh- ‘to be abraded’ < *apa-xrāša- (BD2 p. 275), but without a more certain semantic identification, the etymology must remain be regarded as uncertain.
From the preceding examples we may observe that the only certain examples of initial or medial PIr. *(-)xr- in Bactrian are derivatives of Proto-Iranic *xray- ‘to buy’ (EDIV pp. 446–447 *xraiH- ‘to buy’ < PIE *kʷrei̯h₂- ‘to exchange, acquire through exchange’ LIV2 pp. 395–396). In initial position the reflexes appear to be χιρ- or αχρ-, although, as Sims-Williams notes, since all the vocabulary is derived from the same lexical root, it is possible that some of the nominal forms may have been influenced by the verbal stem. The reflexes exhibiting αχρ- are attested in L and P, dated respectively to Bactrian era 397 (= 601 ce) and Bactrian era 446 (= 668 ce), which are dates after which secondary prothetic α- is normal in χϸονο and therefore we may well expect the initial α- to be secondary from earlier χρ- in these examples also. In medial position *-xr- is stable and preserved without change. I would therefore argue that if a potential Greek source form with initial /xr-/, or /kʰr-/ that was perceived by Bactrian speakers as /xr-/, there is no good reason to expect that it would not have been adopted by Bactrian speakers as /xr-/. Secondly, there is no good evidence for the PIr. cluster *(-)xr- to yield Bact. χϸ- in its inherited vocabulary under any circumstances; this cluster appears to be stable in Bactrian, there is no good reason to expect an ad hoc development /xr-/ > /xš-/ affecting this lexeme alone. For these phonological reasons, a Hellenistic Greek loanword hypothesis seems improbable.Footnote 27
Conclusions
Given the more serious formal difficulties to reconcile the possibility of a Greek loanword with the problematic initial consonant cluster, the need to explain the additional difficulty of assuming a secondary semantic specialisation of ‘time’ > ‘period of time’ > ‘(calendar) year’ within Bactrian may be passed over as unnecessary. I therefore propose that the hypothesis for an origin of Bact. χϸονο from AGk. χρόνος should be rejected. While this conclusion is perhaps a negative one, we may consider that Humbach's original criterion for rejecting an inherited origin was the assumption that ϸ- is the normal Bactrian reflex of PIr. initial xš-, and that more data from the more recently discovered Bactrian documents (cf. §2) suggests that perhaps the development of initial xš- in Bactrian may have been more complicated than originally assumed. Perhaps it may be worthwhile reviving some form of Bailey's earlier proposal (originally for Khot. kṣuṇa-) as a derivative of the PIr. root *xšai- ~ *xšaH- ‘to rule, be lord over’ with a semantic progression ‘(regnal) year’ > ‘(calendar) year’, but for now I leave speculation in that direction a topic for later investigation.
Bibliographical abbreviations used in this article include: BD2 = N. Sims-Williams, Bactrian Documents from Northern Afghanistan II: Letters and Buddhist Texts (Oxford, 2007), EDIV = J. Cheung, Etymological Dictionary of the Iranian Verb, (Leiden, 2007), EWAia = M. Mayrhofer, Etymologisches Wörterbuch des Altindoarischen (Heidelberg, 1992–2001), LIV2 = H. Rix et al., Lexikon der indogermanischen Verben: Die Wurzeln und ihre Primärstammbildungen. Zweite, erweiterte und verbesserte Auflage. (Wiesbaden, 2001), LSJ = H. G. Liddell, R. Scott, H. S. Jones & R. McKenzie A Greek-English Lexicon (9th ed.), With a revised supplement, (Oxford, 1996), NIL = D. Wodtko, C. Schneider & B. Irslinger, Nomina im Indogermanischen Lexikon (Heidelberg, 2008). Linguistic abbreviations used include: AGk. (Ancient Greek), Av. (Avestan), Bact. (Bactrian), Chor. (Choresmian/Khwarazmian) Khot. (Khotanese), Lat. (Latin), MBact. (Manichaean Bactrian), MMP (Manichaean Middle Persian), NPers. (New Persian), OIr. (Old Iranic), OPers. (Old Persian), Parth. (Parthian), PIE (Proto-Indo-European), PIIr. (Proto-Indo-Iranic), PIr. (Proto-Iranic) Prak. (Prakrit), Sogd. (Sogdian), Ved. (Vedic). Citations in this article to the Bactrian Glossary in BD2 have been checked against N. Sims-Williams's unpublished revised version of the glossary originally published in BD2 which includes the vocabulary from documents Nn (N. Sims-Williams, ‘The Bactrian Fragment in Manichaean Script (M 1224)’, in Literarische Stoffe und ihre Gestaltung in mitteliranischer Zeit: Kolloquium anlässlich des 70. Geburtstages von Werner Sundermann, (eds.) D. Durkin-Meisterernst, C. Reck, D. Weber (Wiesbaden, 2009) pp. 245–268), bi (N. Sims-Williams, Bactrian Documents from Northern Afghanistan III: Plates. (Oxford, 2012) p. 21), jj and zd (N. Sims-Williams ‘Two Late Bactrian Documents’ in Coins, Art, and Chronology II: First Millennium C.E. in the Indo-Iranian Borderlands, (eds.) M. Alram, D. Kimburg-Salter, M. Inaba, and M. Pfisterer (Wien, 2010) pp. 203–212).
