You are on page 1of 11

NOTES ON THE ASOKAN ROCK

EDICTS

by K. R. NORMAN

Cambridge

In a footnote to his very valuable study of the Agokan inscriptions, Professor Alsdorf writes I "no systematic, comprehensive investigation of the different versions (apart from the purely linguistic aspect), of their true textual variants, their translation technique, and the mistakes committed in translating has so far been undertaken; it is urgently needed." The following notes, which do not aim at completeness, represent the beginnings of an attempt to find patterns in the committing of mistakes, and in the methods of expression, and to employ the findings to explain certain difficulties. They are written in the hope that they may prove of some value to anyone undertaking the task Alsdorf outlines. 2

1. Confusion of n and y
1.1 Alsdorf has pointed out S that because of the similarity between na and ya in the Brfthmi script it is probable that in RE IX(B) 4 Kpajopaddne is merely a mistake for pajopaddye, and Sh pajupadane goes back to a Br~hmi version showing the same mistake. (For the suggestion that G putraldbhesu hides the same mistake, see para. 6.2 below.) A similar confusion of n and y can be assumed elsewhere. 1 Alsdorf, p. 253. 2 In this article the following abbreviations are used: RE = Rock Edict; PE = Pillar Edict; G = Girnar; K = Kalsi; Dh = Dhauli; J = Jauga.da; M = Mansehra; Sh = Shahbazgarhi; S = Sopara; Y = YerraguOi; Hultzsch = E. Hultzsch, The Inscriptions of Adoka (= CorpusInscriptionurn Indicarum, I) (1925); Bloch = J. Bloch, Les lnscriptions/d'Asoka (1950); Alsdorf = L. Alsdorf, "Contributions to the study of Agoka's inscriptions", BDCRI, XX, pp. 249ff. 8 Alsdorf, p. 262. 4 I follow Hultzsch's division into sections.

NOTES ON THE A~OKAN ROCK EDICTS

161

1.2 (a) In RE X(A) G taddtpano is opposed to K Dh J tadatvdye Y taddtdye Sh M tadatvaye. Bloch suggests ~ reading taddtpane -~ locative of *taddtvana-, although he hints at a possible confusion of n and y. It is, however, most likely that the scribe at G received the incorrect form taddtvdne, which he took to be the eastern nominative singular in -e of a noun with the -tvana- suffix. (For the suggestion that his misunderstanding of this passage was increased by a mistake in the word underlying dighdya, see para. 6.2 below.) (b) In RE XIII(W) K dev(5)na.mpine and dev8nam,pinam,ya in (S) are certainly mistakes for the usual devdna.mpiya-. (c) In RE XIII(X) K s.ayakas.i no vijayas.i is opposed to G sake eva vijaye Sh saya yo vijaye Y sayakasi yeva vijayasi. Although P~li has no in the sense of both negative and emphatic particle, use as the latter in the A~okan inscriptions would be unique here. It is probable that Hultzsch is correct 6 in suggesting that no is a mistake for the emphatic particle yo. (For a more detailed discussion of this particle, see para. 2 below.) 1.3 In RE V(K) G dha.mmayut~nam, is opposed to K Dh -yutdye Sh -yutasa M -yuta. I would suggest that G received the incorrect form -yutdne which was taken to be a genitive plural, the usual equivalence of eastern -e and western -am probably helping this interpretation. From the point of view of establishing the stage at which the error occurred (see para. 7 below), it is clearly essential to find the reading at Y. This however is doubtful, Chakravarty reading -yutdye, and Sircar -yutdnam.. I can make out nothing on the photography

2. yo, ye = eva 2.1 Writing of the usage of particles in the north-western versions, Bloch mentions s Sh vo but does not refer to yo, nor does he distinguish it from the relative pronoun yo in his index, although Hultzsch does list it separately, and refers to its use in his introduction. ~ 2.2 This particle occurs in the following contexts: (a) in RE IV(E) Sh ca yo (after a future tense) is opposed to G Dh ceva K eevd M Y yeva.
5 Bloch, p. 118 n. 1. Hultzsch, p. 46 n. 19.

See Epigraphia Indica XXXII, p. 16.


8 Bloch, p. 82. 9 Hultzsch, p. lxxxv.

162

K.R. NORMAN (b) In RE IV(F) Sh M yo (after a future tense) is opposed to Dh Y

yeva K ceva.

(c) In RE XIII(X) Sh K (if the suggestion in para. 1.2 (c) is correct) yo is opposed to G eva Y yeva. (d) In RE XIII(X) Sh ca yo is opposed to K eva ca Y eva ca. (e) In RE XIV(A) Sh yo seems to be a local addition parallel to vo earlier in the section, where it is opposed to G eva K yevd Y yeva. 2.3 Johansson, followed by Hultzsch, 1~ derived yo < yeva, and if this derivation is correct we might expect to find ye as the intermediate form. Although the occurrence of ye in the required sense has not been noticed hitherto, nevertheless it does exist, but is concealed by faulty sentence division. 2.4 In RE XIII(T) K M ye se Sh yo sa at the beginning of the sentence is difficult to translate, and arouses suspicion. Since the previous sentence ends in K cd Sh M ca, after a verb in the future tense (see para. 2.2 (a) and (b) above) it seems likely that we are to take cd ye, ca yo together. This conjecture is confirmed by Y ceva. We thus have the sought-for particle ye, and another example of Sh yo to add to the occurrences listed above. K Y M se Sh sa is therefore the first word of section (T) and is an example of the "sentence-joining" tat. n 2.5 The emphatic particle ye is presumably to be connected with Pkt je, which in the older texts seems to be used only after infinitives, but is found in Apabhram. ga (continuing an older usage?) in less restricted contexts. Miss Schwarzschild derivesX"je ~ jeva < yeva, but I am not convinced of the correctness of this view. According to this derivation y- was evolved to make the particle more emphatic and expressive, but against this it must be pointed out that this is the only word in the A~okan inscriptions to be so emphasised. We find no forms *yeta, *yedisa-, or *yevam., but the demonstratives eta "here", edisa- "such", and evam. "thus" are all enlarged with h- when it is necessary to emphasise them. I suggest, therefore, t h a t y e v a is to be derived < ye q- eva, rather than that ye and yo are to be derived < yeva. 2.6 We therefore have examples of an eastern form ye with its western counterpart yo, just as we have eastern lz (and P~li) ve with its western
lo Ibid.

n Alsdorf, p. 255 (= p. 254 n. 3).


1~ See L. A. Schwarzschild, "The indeclinable je in Middle Indo-Aryan", Bh~ratiya Vidya, XX-XXI, pp. 211-17. 1~ At Maski (G) ve after heva.m is more likely to be the emphatic particle than the

NOTES ON THE ASOKANROCK EDICTS

163

(and P~li) counterpart vo. The latter may then be merely a dialectal variant, and not a derivative < vd u, as Bloch suggests: 4 2.7 It is possible that the same particle ye is to be seen in the G~ndh~ri Dharmapada where v. 226 has sarvatra ya sapuru.sa vivedi as opposed to P~li Dhammapada (v. 83) sabbattha ve sappurisd vajanti.

3. Confusion of i and
3.1 As Hultzsch has pointed out, 15 i and d are sometimes confuse~l, probably because the Brghmi symbol for d is an incomplete i. We find d written for i as well as i for d, and the mistake is common at G. 3.2 There we have d for i (a) in RE IX(J) na astd opposed to Dh n' athi. (b) in RE XI(C) pitard where mdtari precedes, and a locative form is required. (c) in RE XII(A) vivddhdya opposed to K vividhdye Sh M vividhaye. (d) (For the suggestion that a mistake of d written for i occurred in the word underlying dighdya in RE X(A), see para. 6.2 below.) 3.3 We find i for d at G (a) in RE V(J) ristika- opposed to K Sh M rathika- Dh lat.hika-. (b) in RE X(C) parikamate opposed to K Dh (pa)lakamati Sh M

parakramati.
(c) in RE XI(A) devina.mpriyo for the usual devdnam, priyo. (d) in RE XIII(S) d~ti opposed to K dutd Sh M duta. 3.4 In view of the frequency of this mistake, 16 I have no hesitation in adopting the reading ma(m.)hisara sake at G in RE XIII(X) instead of ma(m. )hd sarasake opposed to K mani.su s.ayaka.si Y raam.nisu sayakasi Sh magi.su spakaspi M ma.ni.su saya. This emendation and word division give a third plural aorist form to match the other versions, with the typical G -r- ending, cf. susu.msera in RE XII(I). 3.5 It is perhaps worth pointing out that in RE V(K) Dh bhat.imayesu opposed to Y bhat.dmayesu (G bhata- K Sh M bhat.a-) may be another example of the confusion between i and d. If bhatd is the correct reading, oblique plural pronoun used as a nominative. 1~ Bloch,p. 82. 15 Hultzsch,p, 56 n. 21. 16 Found also at Rupnath (G) pa for pi, (H) etiya for etffya, (K) lakh- for likh-, (L) -tavffya for -taviya; J sep. edict I eti for eta; PE VII(V) sukhdyan6ya for sukhi-; etc.

164

K.R. NORMAN

it is presumably a nominative plural, agreeing with either dpalam,td or viyapat.d. If the derivation is < bhat.a- we could translate either "my western subjects" or "my servants the maMm~tras". Neither of these translations seems likely, and perhaps we should rather think of a derivation < bhat.t.a- as an honorific title of the mah~mgtras. If we are correct in separating bhat.d and mayesu, then since the other locative plural forms in the sentence refer to classes of mendicants (see para. 6.2 (e) below), it is possible that maya- too is some sort of mendicant. (For the suggestion that mayesu conceals a relative pronoun form, see para. 5.2. (e) below.)

4. Confusion of t and d
4.1 In a number of contexts there is a variation between t and d, in circumstances which seem to indicate that at some stage in the transmission of the edict it was dictated to scribes by a speaker who had a tendency to voice intervocalic 4- > -d-, 17 with the result that his listeners were not always certain of the meaning intended. It is to be noticed that this confusion of t and d is found in the rock edicts only where both spellings make sense, and we may therefore conclude that if the speaker voiced all t sounds, then his listeners corrected them in all contexts where they had no doubt about the meaning. 4.2 (a) In RE V(J) K Sh M hidasukhdye is opposed to Dh Y hitasukhdye, where the scribes were presumably not certain whether hita- "good" or hida "here" was intended. There is, however, never any doubt about hidaloka- or sa(r)valokahita-, where no ambiguity is possible. It should be noted that when the word occurs again in (K) Sh has hita- which perhaps indicates a belated recollection that in some cases d stands for t. (b) In RE VIII(E) G Dh Y tadopayd is opposed to K tatopayd Sh tatopayam. M tatopaya, where doubt could arise as to whether a compound from tad- or tato is intended. (c) (For the suggestion that in RE V(G) K Dh J Y supaddlaye M supadarave there is an example of the voicing of t > d, see para. 6.2 below.) 4.3 Two examples of the voicing of t > d are found at K, but nowhere else, which seems to indicate that they are of local origin. In RE VI (M) K putaddle is opposed to Y putanatdle Sh putra nataro M putra ~7 One of the characteristics of the eastern language of the Buddhist Urkanon (= Old MfigadhT?)according to Lfiders, Beobachtungen, 94-98.

NOTES ON THE ASOKANROCK EDICTS

165

natare G putra prapotrd Dh puta papota. The copy in front of the clerk at K had clearly lost the syllable 18 na- from natdle, and when putatale was read out this was heard as putadale (which made sense) by the scribe. In the same inscription (H) K dose is opposed to some form of tosa- in
all the other versions. Although Liiders 19 regarded this as a mere scribal error, and not an example of voicing, I think that two occurrences of the phenomenon in the same inscription are more than a coincidence, and I would assume that me tose was read out in such a way that it was heard as me dose, which also made sense. 4.4 An interchange of t and d occurs also in the north-western versions but there the reason may lie in a confusion between the Kharo.s.thi symbols for ta and da. In RE VIII(D) Sh ada dhra.mmayatra M ada dhramayada are opposed to G esa dham. mayata K D h Y at~ dham. mayata. Here there is no possibility of the voicing of t > d, since at~ stands for att~, and a doubled intervocalic consonant would not be voiced. (For the suggestion that this is a relative adverb = yatra, see below para. 5.3.) Somewhere during the transmission of the edict Brghmi at~ was transliterated into Kharo.s.thi (for the dependence of the Kharos..thi versions upon Br~hmi originals, see para. 1.1 above), by someone who wrote t badly as d. The scribes at Sh and M copied ada down without understanding it, but Sh was able to correct dhrammayada which was also badly written. We can therefore surmise that Sh and M were here dependent upon the same copy of the edict. In RE IX(N) Sh dhrama.mgalena M dhramagalena opposed to K dham. mamagalena Y dham. mamam. galena shows another example of the two scribes' dependence upon a single copy, and also their ability to follow an obvious fault blindly.

5. Relative pronouns and adverbs without y5.1 In the eastern Pkt underlying the A~okan inscriptions the relative pronoun stern was a- not ya-, and the relative adverb ata (-----yatra). It is not difficult to show that in a number of passages this caused no less confusion to the clerks and scribes, who were faced with the task of interpreting the copy they received before they could translate it into their own dialect, than it has to modern interpreters. z8 Syllablesare also omitted at K in RE IX(D) ka.tavi(ye), RE XII(J) kay~nag<am)d, RE XIII(G) sama(na), RE XIII(Y)palaloki(ki>ye(probably), RE X-IV(C)niky(ay)a.m (as is shown by Y nik~yan.t). 19 Loc. cit., 94.

166

K.R. NORMAN

5.2 (a) In RE II(A) G evam api opposed to Dh J e vd pi K Sh M ye ca Y e ca shows that eva was taken as one word at G = evam, and pi was then written as api. When the same phrase occurred later in the same sentence, the scribe realised his mistake and wrote ye va pi. But he had meanwhile been faced with the problem of am. td which made no sense, and which he had been forced to change to a locative. (b) In RE XIII(H) K e tanatp S h e te.sa M e tan. (cf. Y e tdnam.) are printed as one word by Bloch, on the grounds that e would be an eastern formin the north-west. I think that Bloch was right to raise an objection, 2~ but I think that his reasoning was wrong. I agree with the view of Hultzsch and Liiders that e is a relative pronoun, but I think that the fact that Sh and M did not write ya- proves that they did not recognise it as a relative, but took eta- as the demonstrative pronoun. I would translate "whatever disaster befalls their friends, etc., there, that too is an injury to them." (c) In the same edict there is, I think, another relative pronoun unrecognised by both ancient and modern interpreters. In (G) K Sh M e.sa Y esa is normally taken as a demonstrative pronoun with agabhuti. .su.su.sa etc. It makes better sense, however, if taken as the genitive plural of the relative ( = yes.am), going with te.sam, lower in the same sentence. I would translate "of whom there is obedience etc., to them it is an injury." If this interpretation is correct, then the previous clause must end at vihitd, and since yesu vihitd makes no sense we must divide ye suvihitd and translate "householders who are well-provided for". That this word-division is correct is shown by two facts: (i) Sh and M did not regard yesu as a locative plural, or they would have written ye.su; (ii) ye suvihita is exactly parallel to ye~a.m suvihitdnam, in (H). (d) In (X) in the same edict is another relative pronoun, which modern interpreters supply in their translations, although not recognising its existence. K p u t a p a p o t a me asu Sh M putra papotra m e asu is translated by Hultzsch "sons (and) great-grandsons (who) may be (born) to me" and by Bloch "les ills et petits-fils que je pourrai avoir", but I would suggest that me stands for m' e = me ye. (e) In RE V(K) if mayesu does not refer to a class of mendicants (see para. 3.5 above) then it could perhaps be divided may-e-su. It is not entirely clear what may- could be. It cannot stand for maya, since the eastern form of the instrumental of the first person singular pronoun is me. It is, I suppose, just possible that we have here an example of the so Bloch,p. 127 n. 9.

NOTES ON THE ASOKAN ROCK EDICTS

167

diphthongal pronunciation 21 of me as mai found at Brahmagiri (I), which was heard as may before a vowel. If this is correct, then mayesu is the equivalent of Skt me ye syu.h, and is exactly parallel to m' e asu in para. 5.2 (d) above. 5.3 (a) In RE II(B) and (C) G yata yata K at' ata Dh J at' ata Sh yatra yatra are opposed to M atra atra. There can be no doubt that the original reading was ata ata = yatra yatra, but this was misunderstood at M, although the sense clearly demands a relative adverb. (b) In RE VIII(D) G ten' esa K Dh Y ten' atd Sh M ten' ada is followed by (E) G e t ' ayam. K S Y hera iyam. Dh J tat' esa Sh M atra iya(m. ). The occurrence of the demonstrative adverb heta (with the emphatic h-, see para. 2.5 above) is a clear indication that atd stands for yatra, and we can translate "therefore, where there is a dharmay~tr~, there there is this [i.e. the following occurs]". The version underlying G had taken ata in the sense of atra, and changed it to the more usual eta. The scribe at G took this as the demonstrative pronoun, and changed it to esd to agree with dham.mayfitd. In (E) Dh changed hera to the usual demonstrative tara, while the versions underlying G Sh M dropped h- (as being an eastern vulgarism?). G retained eta, while Sh and M changed it to atra.

6. Variations in vocabulary 6.1 Bloch gives a list ~2 of some of the variations in vocabulary found in the different versions. While there seems to be no obvious reason for some of the variants found, it is possible in other cases to suggest causes for the changes made. 6.2 (a) In RE IX(B) G putraldbhesu is opposed to K pajopaddne Dh J Y pajupaddye S paj@addye Sh pajupadane M prajopadaye. We can assume that the original version had pajopaddye, and there seems to be no reason why G should have changed this. If, however, we assume that G, like K and Sh, received a faulty version showing the y/n variation (see para. 1.1 above), then the scribe might well have interpreted pajopadane in the sense of "taking offspring upon oneself" and glossed it as putra16bha-. The gloss was then incorporated into the actual inscription, as in the next example. (b) In RE X(A) G dighdya is opposed to K ayatiye Dh J dyatiye Y 21 See "Some vowel values in Middle Indo-Aryan', Indian Linguistics, XXI, p. 107. ~ Bloch,p. 83.

168

K.R. NORMAN

dyatiyd Sh M ayatiya, but it is unlikely that dfrgha- can mean "the


future". If, however, we assume that the copy which reached G contained the word *dyatdye showing the d/i mistake (see para. 3.2 above), then the scribe might well have taken this in the sense of "extended, long" and glossed it as digha-. As in the previous example, the gloss then replaced the original. (c) In RE III(C) G niyd- is opposed to K Dh J Y nikham- Sh M nik(r)am-, and the same variant is found in RE VIII(A) and (C). It is possible that the scribe at G was acquainted with the verb ni.skram- in its Buddhist technical sense, and so chose to avoid it. (d) In RE IX(H) G nist.dndya and Dh niphatiyd are opposed to K Y nivutiyd Sh M nivut.iya. It is not impossible that the scribes at G and Dh both mistook nivuti- for the Buddhist technical term (cf. P~li nibbuti-) and therefore avoided it. In fact, as the verb in (J) and (M) shows (K Sh M nivat.- Y nivat-), nivuti- is < nirv.rtti-, not nirv.rti-. (e) In RE IV(C) G thairasusrusd is opposed to Dh Y vu.dha- Sh vu.dhanam. M vudhrana. In RE V(L) G thairesu is opposed to K Dh Y mahdlake Sh M mahalake, and we may assume that in (K) too G had thairesu opposed to K vudhesu Y vud.hesu Sh vu.dhe.su M vudhre.su, but Dh mahdlakesu. In RE VIII(E) G thairdnam, is opposed to K vudhdnam. Dh J S Y vu.dhdna.m Sh vu.dhanam M vudhana. In RE V there seems to be a clear distinction between vud.ha- and mahdlaka-, and I would accept the translation "old" for the latter. But since there is some evidence for thinking that the ibhyd were identical with the ~jivikas and the andthd with the Nirgranthas, ~s the v.rddhd are presumably the ndndpdsam..dd mentioned in conjunction with Brahman.as, Ajivikas, and Nirgranthas in PE VII(Z). We should therefore translate "elders, venerable ones", and vu.dhasus~sd would be synonymous with galus.u.sus.d in RE XIII(G). It is unlikely that vudha- means the member of any particular sect, 24 but if it is correct to think that the scribe at G was acquainted with Buddhism, then it could well be that in his view the only true elders, or truly venerable ones, were the Buddhist theras, and he therefore wrote thaira-. In RE V he was confused by the juxtaposition of vu.dha- and mahdlaka-, and wrote thaira- in both places, just as the scribe at Dh, equally confused, wrote mahdlaka- in both places. (f) In RE V(G) G Sh sukaram, is opposed to K Dh J Y supaddlaye M supadarave. Modern interpreters are not agreed about the meaning of 23 I hope to write about this at greater length elsewhere. 24 In Skt v.rddha-is found in the sense of "religiousmendicant", see Monier-Williams, Skt Dict., s.v.

NOTES ON THE ASOKAN ROCK EDICTS

169

supaddraya-, and from the way in which the scribe at M made the usual
change of l ~ r but did not change v > y, it seems likely that he, and possibly the other scribes too, did not understand the word. This probably explains why the clerk whose version underlies G and Sh changed the word to sukara.m. What we do not know is whether he understood supaddlaya- when he interpreted it in this way, or whether he assumed the meaning from the context, extracting su- from supaddlayaand kara- from -kat.am. kdsati. That both G and Sh go back to one original rather than being independent guesses is shown by the fact that they both omit ndma. In view of the evidence given above (para. 4) for the voicing of t > d, I would suggest that supaddlaya- stands for *supratdraya- with the meaning "easy to extend, easy to spread". 6.3 Some apparent variations in vocabulary are merely mistakes: (a) In RE IX(B) G ucdvacam, opposed to bahu(kam.) in the other versions is due to the scribe inadvertently repeating the beginning of the sentence. (b) In RE XII(E) G prakaran, ena is opposed to K Y dkdlena Sh M akarena. Although Hultzsch ("in every case"), and Bloch ("~t chaque occasion") both translate prakara.na-, there can be no doubt that this represents a slip of the clerk's tongue or pen for dkdrena, tena tena prakara.nena being evolved because of tamhi tamhiprakaran, e just before.

7. Conclusions
7.1 Consideration of the various mistakes and methods of expression enable us to conclude that the chain of transmission of the rock edicts usually took the following form. 7.2 (a) The king's commands were dictated by someone whose pronunciation showed some M~gadhan peculiarities (e.g. d for t ~5) to one or more clerks. They wrote down in Br~hmT what they heard, correcting any mispronunciations, which they presumably regarded as vulgarisms, except where the phrase was ambiguous (e.g. hita-/hida-) or the meaning unknown to them (e.g. supaddlaya-). Mistakes were made (e.g. n written for y), and when there was more than one scribe differences began to arise (e.g. e ca opposed to e vdpi in RE II(A)). (b) As many copies as were required for the regions were made from the original(s). It seems clear from the close similarity between the versions at Dh and J that they both used the same copy, and Sh and M sometimes had only one copy between them (as shown by ada and

170

K.R. NORMAN

dham. mam.gala-), but generally there was no consistency in the arrangement of which regional copy was made from which original. So we find that in RE II G Dh J follow the version with e vd pi, but in RE V G Sh follow the version with sukaram, and aparigodhdye. Besides the mistakes which inevitably crept into individual regional copies at this stage (e.g. *dyatdye for dyatiye, *putatdle for putanatdle), some dialect changes were also introduced (e.g. eta written for esa in RE VI(J) in the version sent to Sh, which the local scribe later changed to etra). (c) The versions intended for Sh and M were transliterated into Kharo.st.hi. (d) Translation into the local dialect was carried out in the various regions, to the best of the clerk's or scribe's ability. ~6 In some cases, at least, the translation seems to have been made as the carving was in progress. So in RE VIII(E) the scribe at G copied from his original dasa~e ca ddne ca... dasan, e ca, presumably taking these forms as locatives. Reading on to the end of the sentence he found dham. mdnusathi ca dham.mapalipuchd, and realised that the preceding words too were nominatives. He accordingly continued hiram, n.apatividhdno ca janapadasa ... darsanam. , but did not correct what he had already inscribed. If the translation had been made before the carving was begun, there seems to be no good reason for such half-corrected misinterpretations, which are numerous. That the edicts as we have them are sometimes eye-copies is shown by the typical copying errors which are found (e.g. in RE XIV(E) Y sa.mkhdsam.khaydya for sam.khaye). That the version was sometimes dictated to the scribe by a clerk is shown by mistakes which seem more likely to be aural than visual (e.g. K dose and putadale in RE VI, and the phonological variations and changes in word order in the two versions of RE VI(E) and (F) at Sh).
7.3 The king's original orders were, therefore, inscribed as they had been misheard, miscopied, and misinterpreted, to a greater or less extent, by the various members of the secretariate. It must be our aim to remove the faults which have crept in (of which the examples given above are only a small selection), and try to find out exactly what A~oka said.

25 There is also some evidence for k > g. 26 It is most improbable that the language of S resembled the true dialect of the region. Similarly, the variations between the language of RE and Minor RE at Y represent a change of translator rather than a genuine development of language.

You might also like