Pramāṇa · Search · About · API

Early Tamil Epigraphy: From the Earliest Times to the Sixth Century A.D. c21: 2. Decipherment

Early Tamil Epigraphy: From the Earliest Times to the Sixth Century A.D. c21

Language
English

Contents

2. Decipherment

2.1 Introduction

c21.s1.1

The term 'decipherment' is generally employed to denote the successful reading and understanding of an unknown script in a language which is, to begin with, unknown. It may, therefore, appear *prima facie* inapt to apply the term to the reading and interpretation of the earliest cave inscriptions in the Tamil country, as neither the script (Brāhmī) nor the languages likely to be involved (Prakrit and Tamil) can be described as 'unknown'. However, the fact remains that when the cave inscriptions were discovered, they were found to be wholly unintelligible, and it took the combined labours of several scholars for well over half a century before the inscriptions could be understood.

c21.s1.2

The following are the main stages by which the cave inscriptions became progressively more intelligible.

c21.s1.3

(1) Determination of the true phonological character of the script, which ruled out Indo-Aryan as the language of the inscriptions.¹

c21.s1.4

¹ K.V. Subrahmanya Aiyer 1924.

c21.s1.5

(2) Identification of the special Tamil characters *ḻ*, *ḷ*, *ṟ* and *ṉ* in the script.²

c21.s1.6

² *ḻ* by H. Krishna Sastri (1919) and the rest by K.V. Subrahmanya Aiyer (1924).

c21.s1.7

(3) Discovery that the cave inscriptions followed orthographic conventions different from those of the Mauryan Brāhmī, but related to the Bhattiprolu casket inscriptions.³

c21.s1.8

³ T.N. Subramanian 1938 and 1957. T.V. Mahalingam 1967. I. Mahadevan 1968 and 1971.

c21.s1.9

(4) Study of the cave inscriptions *in situ* and editing them direct from the stone.⁴

c21.s1.10

⁴ I. Mahadevan, *Corpus of the Tamil-Brāhmī Inscriptions 1966* (1968) with further improvements in this volume.

c21.s1.11

The whole process can be truly described as 'decipherment' as it revolutionised our understanding of the script and the language of the inscriptions and led to fresh insight into the origin and evolution of Tamil writing systems.

c21.s1.12

This chapter is a brief chronological account of the decipherment of the Tamil-Brāhmī script. The results of the decipherment relating to the language and contents of the inscriptions are presented respectively in the next two chapters.

2.2 Venkayya's reading

c21.s2.1

Venkayya was the first to identify the script of the cave inscriptions as Brāhmī.⁵ His judgement that "the alphabet of the inscription(s) resembles that of the Asoka edicts and may be assigned roughly to the end of the 3rd and beginning of the 2nd century B.C."⁶ has stood the test of time. However, his assumptions, even before any of the inscriptions could be read, that "the language employed in them is Pāli" and that "it was understood in the Pāṇḍya country even at that early period"¹ proved to be incorrect and stalled his efforts to understand the inscriptions. He attempted to read the inscription on a stone bed at Mettuppatti² as follows:

c21.s2.2

⁵ *ARE 1906-07*: paragraphs 1-5; *1907-08*: paragraphs 1-7.

c21.s2.3

⁶ *ARE 1906-07*: paragraph 3.

c21.s2.4

¹ *ARE 1906-07*: paragraph 3.

c21.s2.5

² *ARE 45a / 1908* = No. *25*.

c21.s2.6

The writing on the first bed seems to be *anatai ariyā*. If my reading be correct, one would be tempted to take *anatai*, which occurs also in three other beds, as derivative of the Vedic root *an* 'to live'. If this be the correct derivation of *anatai*, it might mean *living*. The second word in the first bed is *ariyā* which is, perhaps, meant for *ariyānā* 'of the (Buddhist) saints' (?).³

c21.s2.7

³ *ARE 1907-08*: paragraph 7.

c21.s2.8

This very first attempt to read a cave inscription illustrates a major difficulty, namely, that even if the characters are read correctly with values as in the Brāhmī script, the text turns out to be unintelligible. It is interesting that Venkayya did not consider the possibility that the text, or at least part of it, could be in Tamil, the local language.⁴ In any case, Venkayya did not take his own reading too seriously and admitted that "we must wait for more allied inscriptions in good preservation before we can be sure about their interpretation".⁵

c21.s2.9

⁴ According to our present understanding, the inscription reads *antai ariyti* (the last letter *ti* being engraved on the next bed), where *antai* is a Tamil honorific and *ariyti* (< Pkt. *hariti*) a personal name. See Commentary.

c21.s2.10

⁵ *ARE 1907-08*: paragraph 7.

2.3 Krishna Sastri's readings

c21.s3.1

Krishna Sastri made the next attempt to read the cave inscription at Sittannavasal,⁶ but admitted that he was not able to suggest any possible interpretation of the inscription.⁷ However, his tentative reading illustrates another characteristic of the Tamil cave inscriptions, namely, that even where the writing is bold, clear and exceptionally well-preserved as in this case, the text is almost wholly unintelligible. We know now that Krishna Sastri could not have succeeded in reading the inscription as the clues to a correct understanding of the script were yet to be discovered. However, Krishna Sastri had already started moving away from Venkayya's position and, adopting a more cautious approach, declared that the language of the cave inscriptions was "neither Pāli nor Dravidian".⁸

c21.s3.2

⁶ *ARE 388A / 1914* = No. *49*.

c21.s3.3

⁷ *ARE 1914-15*: paragraph 2. See section 1.3.2 and Commentary (No. *49*) for interpretation of this inscription.

c21.s3.4

⁸ *ARE 1909-10*: paragraph 5.

c21.s3.5

Krishna Sastri lamented that even though the Brāhmī cave inscriptions were brought to the notice of scholars and were also submitted to some for critical study, they still remained "uninterpreted" and that scholars "have not paid to these ancient records the attention they deserve".⁹ Probably in desperation, he suggested to Ramaprasad Chanda, described as a brilliant scholar in ethnology "to take up these curious Brāhmī documents of Southern India for study and make an attempt, however slight it may be, to interpret them".¹ He took Chanda on a tour of the caves and supplied him with transcriptions and estampages of the inscriptions. However, his hope that Chanda was "shortly going to publish a monograph on these inscriptions which is expected to throw some fresh light on these hitherto obscure documents" remained unfulfilled.² Nothing more was heard of the project which was in any case doomed from the start, as Chanda, a scholar from Bengal and presumably unfamiliar with Dravidian, was not the best choice for the job.

c21.s3.6

⁹ *ARE 1917-18*: paragraph 8. At least part of the blame for this unhappy situation lay with the Government Epigraphist who published photographs of the estampages in such small size as to be virtually unreadable. *ARE 1911-12* has one Plate with 12 inscriptions and *ARE 1917-18* has another with 22 inscriptions!

c21.s3.7

¹ *ARE 1917-18*: paragraph 8.

c21.s3.8

² Ibid. paragraph 10.

c21.s3.9

The first major study of the cave inscriptions is the paper presented by Krishna Sastri at the First Oriental Conference held in Poona (Pune).³ The paper includes his readings of 31 inscriptions⁴ from 11 sites. He did not translate or interpret the texts and confined himself to commenting on some of the words he could read in the inscriptions. Notwithstanding this serious limitation, Krishna Sastri's paper represents important, even if tentative, advances in understanding the script and the language of the cave inscriptions.

c21.s3.10

³ H. Krishna Sastri 1919: pp. 327-348 (= HKS).

c21.s3.11

⁴ 32 inscriptions according to the numbering in the present Corpus.

2.3.1 Script:

c21.s3.t1.1

Krishna Sastri noted for the first time the "purely southern characteristic" of the palaeography of the cave inscriptions comparing them with those of the Brāhmī inscriptions of Ceylon (Sri Lanka).⁵ The comparison was fruitful as it enabled him to identify the occurrence of the Dravidian *ḻ* in the Tamil cave inscriptions based on the earlier discovery of this character by Parker in the Brāhmī inscriptions of Sri Lanka.⁶ From the same source, he also identified *ī* which occurs earlier in the Tamil cave inscriptions when compared with the Brāhmī inscriptions elsewhere in the country.⁷ This was a valuable contribution by Krishna Sastri to the study of the Tamil-Brāhmī script. He also noticed the occurrence of three other unusual characters in the script which, however, he could not identify.⁸ Later, Subrahmanya Aiyer correctly identified them as the special Tamil characters for *l*, *r* and *ṉ*.⁹

c21.s3.t1.2

⁵ H. Krishna Sastri 1919: pp. 331-332, 347-348.

c21.s3.t1.3

⁶ Ibid. p. 334, n.4; pp. 347-348. H.W. Parker 1909: Nos. 26 & 27 at p. 429; No. 38 at p. 431.

c21.s3.t1.4

⁷ H. Krishna Sastri 1919: p. 348.

c21.s3.t1.5

⁸ Ibid.

c21.s3.t1.6

⁹ See section 2.4.2.

2.3.2 Orthography:

c21.s3.t2.1

Krishna Sastri was the first to compare the Tamil cave inscriptions with the Bhattiprolu casket inscriptions of Andhra.¹⁰ However, he did not correctly apply the rules of the Bhattiprolu orthography discovered earlier by Bühler.¹¹ He assumed that a character in the cave inscriptions, which resembled *nā* of the casket inscriptions, was of the "Bhattiprolu type".¹ This led him to the implausible conclusion that "the numerous words ending in *nā* in all the inscriptions may be taken to be the genitive plurals ending in *-nāṃ*",² resulting in a gross exaggeration of the "Prakrit element" in the cave inscriptions. Actually, the Bhattiprolu *nā* (or, more generally, the *-ā* medial sign of Bhattiprolu) never occurs in the Tamil-Brāhmī script. It was left to Subrahmanya Aiyer to correct the reading of *nā* to *ṉ* (the alveolar nasal which occurs only in the Tamil-Brāhmī script).³

c21.s3.t2.2

¹⁰ H. Krishna Sastri 1919: p. 347. See also section 5.6.3 on the Bhattiprolu casket inscriptions.

c21.s3.t2.3

¹¹ See section 6.6 for discussion on the orthography of the Bhattiprolu inscriptions.

c21.s3.t2.4

¹ H. Krishna Sastri 1919: p. 337, n.11 and p. 348.

c21.s3.t2.5

² H. Krishna Sastri 1919: p. 347.

c21.s3.t2.6

³ See section 2.4.2.

c21.s3.t2.7

Krishna Sastri was again the first to guess correctly (though rather intuitively from his knowledge of Tamil epigraphy) that at least some of the consonants in the cave inscriptions have to be treated as basic.⁴ Actually, this is one of the main features of the Bhattiprolu orthography and it applies almost without exception to all the unmarked consonants in the early cave inscriptions.

c21.s3.t2.8

⁴ H. Krishna Sastri 1919: p. 347.

2.3.3 Language:

c21.s3.t3.1

Krishna Sastri was the first to recognise the presence of the "Dravidian element" in the cave inscriptions. He could, however, notice only four or five words "apparently in Tamil" (e.g., *ūr*, *nāṭu*) and the "superfluous *y* after words ending in medial *-ai*" as part of the Dravidian element.⁵ On the other hand, the Prakrit element was, according him, " more or less clearly expressed" in the inscriptions.⁶

c21.s3.t3.2

⁵ Ibid. p. 346.

c21.s3.t3.3

⁶ Ibid. p. 347. For our present understanding of the true proportions of Dravidian and Indo-Aryan elements in the cave inscriptions, see sections 3.1.5.

c21.s3.t3.4

Apart from his failure to understand the orthographic peculiarities of the inscriptions and to identify the special Tamil characters for *ḷ*, *ṟ* and *ṉ*, Krishna Sastri's readings have also been affected by his reliance on the unsatisfactory estampages of weather-worn inscriptions.⁷

c21.s3.t3.5

⁷ There is internal evidence in his 1919 paper indicating his reliance on estampages and not on direct reading from the stone. His mixing up the estampages at Mangulam has already been referred to in section 1.2.1(ii). The sequence of his readings at Kongarpuliyankulam and Mettuppatti is also at variance with the order *in situ* (See Nos. *12* & *13* and *28* & *29*.) Direct reading from the stone would have produced more accurate texts in other cases also.

2.4 Breakthrough: Subrahmanya Aiyer's decipherment

c21.s4.1

The paper by Subrahmanya Aiyer presented at the Third Oriental Conference held at Madras (Chennai) is the breakthrough leading to an understanding of the real nature of the script and language of the Tamil cave inscriptions.⁸ In this paper, he presented his readings and interpretations of 30 inscriptions from 12 sites. While his readings of individual inscriptions are not without problems as we shall see presently, his elucidation of the phonological structure of the script, crucial identification of the special characters for the Tamil sounds *ḻ*, *ṟ* and *ṉ*, and his demonstration that the language of the cave inscriptions is basically Old Tamil with an admixture of Prakrit loanwords have remained valid and constitute the real point of departure for all further work in the field.

c21.s4.2

⁸ K.V. Subrahmanya Aiyer 1924: pp. 275-300 (=KVS).

2.4.1 Phonological structure of the script

c21.s4.t1.1

Subrahmanya Aiyer begins by pointing to a "powerful misguiding factor" that what is written in Brāhmī must necessarily be in Pāli or Sanskrit, as the script and the language are "inseparably associated in our minds".¹ Having freed himself from this assumption, he proceeds to delineate the phonological structure of the script employed in the cave inscriptions as follows.²

c21.s4.t1.2

¹ K.V. Subrahmanya Aiyer 1924: p. 282.

c21.s4.t1.3

² Ibid. pp. 283-284.

c21.s4.t1.4

(1) The *varga-prathamas* have generally been used.
(2) The soft consonants i.e., the *varga-tṛtīyas* are conspicuous by their absence.
(3) *śa* and *ṣha* are not met with, though *sa* is occasionally found.
(4) The lingual *ḷa* occurs frequently.
(5) Almost all the vowels with the exception of *ai*, *au*, *ṛi*, *lṛ*, *aṁ* and *aḥ* are represented.
(6) In the case of combined consonants the occurrence of short *e* and *o* deserve special attention, the two being the special characteristics of the Dravidian alphabets.³
(7) The aspirates are seldom used; the only two letters that are met with are *ṭha*⁴ and *dha*.
(8) Some symbols which never find use in the contemporary Asokan edicts are here employed and the sounds which they represent remain to be determined. This is a factor which at once points out that the language employed in the inscriptions contains sounds that could not be represented by the symbols extant in the Asokan code.

c21.s4.t1.5

³ The reference is obviously to the short *e* and *o* sounds in Dravidian, which are not, however, graphically represented in the early cave inscriptions.

c21.s4.t1.6

⁴ *ṭh* does not in fact occur. The earlier form of *ḻ* was misread as *ṭh* by Krishna Sastri and Subrahmanya Aiyer.

c21.s4.t1.7

Subrahmanya Aiyer also drew pointed attention to the complete absence of conjunct consonants in this script.⁵ His analysis decisively ruled out Indo-Aryan as the language of the cave inscriptions.

c21.s4.t1.8

⁵ Ibid. p. 288.

2.4.2 Discovery of the special Tamil characters

c21.s4.t2.1

***ḻ*: The letter *ḻ* was first identified in the cave inscription at Tirupparankunram.**⁶ Krishna Sastri commented that "one can venture to say that it is somewhat like the modern Tamil letter *ḻa*".⁷ However, he did not venture far enough and finally opted to read it as *jaṁ* on the basis of its "very slight" similarity to the Asokan *j* at Khalsi.¹ Subrahmanya Aiyer identified the letter as *l* on the basis of its remarkable resemblance to the modern Tamil form.² As he pointed out, its significance would not have been missed but for its association with Brāhmī. The identification suited the context as it produced a recognisable Tamil word (*īla-*). The correctness of the identification has been amply confirmed by later discoveries of the letter in appropriate linguistic context in stone and pottery inscriptions.

c21.s4.t2.2

⁶ Corpus: No. 55.

c21.s4.t2.3

⁷ H. Krishna Sastri 1919: p. 336, n.7.

c21.s4.t2.4

¹ H. Krishna Sastri 1919: p. 336, n.7.

c21.s4.t2.5

² K.V. Subrahmanya Aiyer 1924 : pp. 284-285.

c21.s4.t2.6

However, even Subrahmanya Aiyer could not identify the still earlier form of the letter *l* with the 'circular top' found at Mangulam and other early sites, and misread most of the occurrences as *ṭh*.³ It was only in 1946 that the earlier form was correctly identified and illustrated in the palaeographic chart published in the report on the excavation at Arikamedu.⁴

c21.s4.t2.7

³ Cf. Nos. *2*, *3*, *34*, *43* & *44* for the older forms of *l*, which were not correctly identified in K.V. Subrahmanya Aiyer 1924.

c21.s4.t2.8

⁴ R.E.M. Wheeler *et al.* 1946: Fig. 45 at p. 110. Subrahmanya Aiyer was one of the experts contributing to this section in Wheeler's report. The following comment is made on this letter in the report (p. 111): "The letter *la* has a more circular form in the cavern inscriptions, whilst at Arikamedu it has the almost modern open form. This is a point in favour of a later date for the Arikamedu graffiti".

c21.s4.t2.9

***r*: Krishna Sastri noticed the occurrence of two unusual letters in the cave inscription at Anaimalai,⁵ "the formation of which is not found elsewhere". He attempted to read them as *ḍu* and *ḍai*, but admitted that the readings were not certain and wondered whether they might be "conjunct consonants".⁶ Subrahmanya Aiyer was able to identify the letter correctly as *r*.⁷ He inferred from the shape of the letter resembling "a badly shaped *ṭu*" that it must be allied in sound with *ṭ*. He drew attention to the addition of medial signs to the letter, which showed that it must be a simple consonant and not a 'conjunct consonant' as suggested by Krishna Sastri. He pointed out that the only letter allied to *ṭ* for which a new character had to be added to the Asokan Brāhmī was the Dravidian *r*. He located the clinching evidence in the cluster *-nr-* occurring in the word *kunra-* 'hill' in the Anaimalai inscription itself, which conclusively established the value of the letter in question as *r*.⁸ Subsequent discoveries of the letter in stone and pottery inscriptions have fully confirmed the correctness of the identification.

c21.s4.t2.10

⁵ Corpus: No. *60*.

c21.s4.t2.11

⁶ H. Krishna Sastri 1919: 334, n.3.

c21.s4.t2.12

⁷ K.V. Subrahmanya Aiyer 1924: p. 285.

c21.s4.t2.13

⁸ Ibid. pp. 294-295.

c21.s4.t2.14

However, even Subrahmanya Aiyer was not always able to distinguish between *r* and *ṭu* in the earlier inscriptions. The letter *r* occurs ten times in the inscriptions read by him; he was able to identify the letter correctly only in four cases and misread the other occurrences mostly as *ṭu*.⁹

c21.s4.t2.15

⁹ The letter *r* was correctly identified by Subrahmanya Aiyer in Nos. *11* (once), *49* (once), and *60* (twice), but not in Nos. *3*, *6*, *11* (once), *12* (twice) and *49* (once).

c21.s4.t2.16

***ṉ***: The most frequent letter in the Tamil cave inscriptions resembles an inverted J. It does not occur in the Brāhmī script, but looks somewhat like the *nā* of the Bhattiprolu script, which misled Krishna Sastri to identify it as *nā* of the "Bhattiprolu type".¹ Subrahmanya Aiyer took a different line and was successful in identifying the letter as the alveolar nasal *n* in Tamil. It is instructive to follow his line of reasoning summarised below.²

c21.s4.t2.17

¹ H. Krishna Sastri 1919: p. 337, n. 11 and p. 348.

c21.s4.t2.18

² K.V. Subrahmanya Aiyer 1924: p. 284.

c21.s4.t2.19

(i) The letter is the most frequent one in the cave inscriptions and generally occurs in word-final position.

c21.s4.t2.20

(ii) The letter cannot be *nā* which also occurs in the same inscription along with the new letter.

c21.s4.t2.21

(iii) The only letter in Tamil which is mostly word-final and is allied in sound to *n* is *n*.

c21.s4.t2.22

(iv) The value *n* fits the context in all cases (see especially *kunra-* 'hill' in the Anaimalai inscription).

c21.s4.t2.23

It is no exaggeration to claim that Subrahmanya Aiyer's identification of the correct values of the special Tamil characters for *ḷ*, *r* and *n* is a feat which places him in the front rank of palaeographers along with Prinsep and Bühler.

2.4.3 Identification of Tamil grammatical elements

c21.s4.t3.1

Subrahmanya Aiyer was the first to demonstrate convincingly the presence of grammatical features of the Tamil language in the cave inscriptions. Among the features pointed out by him are the following.³

c21.s4.t3.2

³ Ibid. pp. 286-287 & 296.

c21.s4.t3.3

(i) Pronominal terminations *-an*, *-ān*, *-ōn* and *-ōr*.

c21.s4.t3.4

(ii) Neuter nominative suffix *-m*.

c21.s4.t3.5

(iii) The archaic genitive case-ending *-a*.

c21.s4.t3.6

(iv) Verb forms like *kuṭupita*, *ceyta*, *piranta*, etc.

2.4.4 Language of cave inscriptions

c21.s4.t4.1

Subrahmanya Aiyer's analysis not only ruled out Indo-Aryan as the language of the cave inscriptions, but also established convincingly that the language is Old Tamil with an admixture of Prakrit loanwords. His conclusion that "it is quite consistent that in a Tamil country, the inscriptions, meant to be understood by the race, which is not known to have spoken any other language must be Tamil"⁴ has never been seriously disputed thereafter.

c21.s4.t4.2

⁴ Ibid. p. 286.

2.4.5 Problems with Subrahmanya Aiyer's readings

c21.s4.t5.1

A study of Subrahmanya Aiyer's work reveals that in spite of the correct premises he proceeded from, his readings were often less than successful mainly on account of the following reasons.

c21.s4.t5.2

(a) *Incorrect orthography

c21.s4.t5.3

Subrahmanya Aiyer was aware that the orthographic conventions relating to medial vowels in the Tamil cave inscriptions were different from those of the Asokan Brāhmī script. He refers to the Bhattiprolu casket inscriptions and actually cites Bühler's crucial discovery that "the medial short *-a* is indicated in the Southern Mauryan script by the addition of a horizontal stroke to the top of the consonant, which denotes the long *-ā* in Mauryan", but rejects its application to the cave inscriptions on the ground that "the addition of the horizontal stroke without any further change represents the long *-ā* in the Pāṇḍyan inscriptions".¹ Subrahmanya Aiyer's alternative suggestion that there is "unnecessary lengthening of penultimate short consonants (read syllables) in some words"² is not supported even by his own readings; e.g., *sālākāṇa* > *sālakaṇ*, where the last two vowels are 'lengthened'. The simpler solution that, unlike at Bhattiprolu, the medial stroke represented both *-a* and *-ā* according to the context in the cave inscriptions did not occur to him.

c21.s4.t5.4

¹ K.V. Subrahmanya Aiyer 1924: p. 283.

c21.s4.t5.5

² Ibid.p. 286.

c21.s4.t5.6

(b) *Overestimating the Prakrit element

c21.s4.t5.7

According to Subrahmanya Aiyer, "a large number of Sanskrit words in their Prākṛit forms" occur in these records.³ Later studies have shown that he tended to overestimate the presence of the Indo-Aryan element in these records, though not perhaps to the same extent as was done by Krishna Sastri. Many of the so-called Prakrit words turn out to be due to incorrect reading. For example, out of the fifteen words listed by Subrahmanya Aiyer in their Sanskrit form to illustrate his point, about half the number do not in fact occur in the inscriptions read by him.⁴

c21.s4.t5.8

³ Ibid.

c21.s4.t5.9

⁴ Ibid. *karaṇi*, *kārita*, *kula*, *caitya*, *putra*, *yaksha* and *vatsa* in this list are the result of misread words.

c21.s4.t5.10

(c) *Reliance on unsatisfactory estampages

c21.s4.t5.11

Even though Subrahmanya Aiyer discovered many of the cave inscriptions and had seen most others *in situ*, he appears to have edited the inscriptions for publication relying on unsatisfactory estampages of weather-worn inscriptions. Otherwise, the mixing up of estampages at Mangulam and the incorrect sequence of reading the inscriptions at Kongarpuliyankulam, Vikkiramangalam and Mettuppatti could not have happened.⁵ I have no doubt that direct editing from the stone would have produced more accurate texts in most cases.

c21.s4.t5.12

⁵ See notes on locus in Nos. **3**, **12**, **13**, **21**, **27** and **28**. K.V. Subrahmanya Aiyer himself states that his interpretations are based "mostly on the three plates given by Mr. Krishna Sastri" (1924: p. 282).

c21.s4.t5.13

(d) *Some 'strange' readings

c21.s4.t5.14

The cumulative effect of the problems discussed above has been to produce some strange readings as the examples given below show. (The upper lines in normal type reproduce Subrahmanya Aiyer's readings, while the lower lines in italics are the corresponding texts in the present Corpus).

c21.s4.t5.15

(1) **veḷ aḍaiy nikāmatāko potir yakāsīti kāaritāva . . .**¹
***veḷ-aṟai-y nikamatu kāviti-iy kāḷitika antai . . .***

(2) **ūpācāa pōta neṭula vōccō koṭu paḷiī.**²
***upaca-aṉ toṇṭi=lavōṉ koṭu paḷi-i.***

(3) **kuṭu koṭālaku itātāviṇ ceṭu atāṉ lēṇ.**³
***kuṟu koṭalku ittavaṉ ceṟ-ataṉ-ōṉ.***

c21.s4.t5.16

¹ K.V. Subrahmanya Aiyer 1924: p. 292: No. IVC= No. *3*.

c21.s4.t5.17

² Ibid. p. 294: No. V=No. *10*.

c21.s4.t5.18

³ Ibid. p. 295: No. VIIIC=No. *12*.

2.4.6 Assessment of Subrahmanya Aiyer's work

c21.s4.t6.1

Notwithstanding the errors in Subrahmanya Aiyer's readings of individual inscriptions, which have been corrected by later research, his basic findings on the phonological structure of the script and nature of the language of the cave inscriptions have remained intact. They have provided the secure foundation for further work which has revealed that the cave inscriptions turn out to be in simple and intelligible Tamil when read with proper values for characters and with appropriate orthographic conventions.

2.5 A comic interlude

c21.s5.1

Fifteen years after Subrahmanya Aiyer's path-breaking study appeared, Narayana Rao made an unsuccessful effort to put the clock back.⁴ Basing himself wholly on Krishna Sastri's readings, Narayana Rao argued that since all known inscriptions in the Brāhmī script were in Prakrit, the cave inscriptions in this script in the Tamil country must also be in the same language. He suggested that these inscriptions were in "a form of Prākṛit described by the Prākṛit grammarians as Paiśācī".⁵ According to him, "the Pāṇḍya country ... is a tract where the prevailing language is Paiśācī".⁶ To prove his thesis, he tortured Krishna Sastri's readings to make them yield texts supposedly in Prakrit, getting his clue from the *-nā* (*>-nāṁ*) ending suggested by Krishna Sastri for many of the inscriptions. For example, the long inscription at Mangulam read by Krishna Sastri and Subrahmanya Aiyer becomes (in the Sanskritised version by Narayana Rao):⁷

c21.s5.2

⁴ C. Narayana Rao 1938-39: pp. 362-376 (=CNR).

c21.s5.3

⁵ Ibid. p. 376.

c21.s5.4

⁶ Ibid.

c21.s5.5

⁷ Ibid. No. IV A at p. 368.

c21.s5.6

*gaṇakānām nāthā(nām) Śrī-yakshāṇām dharmam;*
*idha (itaḥ) nartinām sārthavāhakānām*
*Simhaḷānām śrēshṭhikānām dāna-dēyam;*
*śrēshṭhikānām caitya-pālikā.*

c21.s5.7

Narayana Rao's criticism of Subrahmanya Aiyer's work can, with better justification, be applied to his own readings:¹

c21.s5.8

¹ C. Narayana Rao 1938-39: p. 376.

c21.s5.9

Mr. Subrahmanya Aiyar made a mess of the whole lot of the inscriptions acting on the wrong lead given by Mr. Krishna Sastri concerning the existence of Tamil words. For his purpose of making them appear as complete Tamil records, he took undue and unwarranted liberty with them; he cut off vowels as well as consonants, added others, twisted some more, made wrong grouping of the letters, tried to force meanings into the words he had created, and when he failed to draw any sense out of them, as very often he could not draw, he was satisfied with saying that they are all proper names of persons or places.

c21.s5.10

The unsound attempt by Narayana Rao was not taken seriously by anyone else and it has now sunk into wholly deserved oblivion.

2.6 Elements of Bhattiprolu orthography in cave inscriptions

c21.s6.1

As we saw earlier, though Krishna Sastri and Subrahmanya Aiyer refer to the Bhattiprolu casket inscriptions and their unique orthographic conventions governing the depiction of medial vowels, neither of them really understood how these were related to the Tamil cave inscriptions. The next major advance in this direction is due to Subramanian who authored the first book in Tamil on palaeography in 1938 with an English translation by himself in 1957.² Subramanian did not read or interpret individual inscriptions and confined himself to a general discussion on the palaeography of the script illustrating it with the first published palaeographic chart based on the cave inscriptions discovered up to that time.³ Subramanian was the first to recognise and state clearly that the Bhattiprolu conventions of

c21.s6.2

² T.N. Subramanian 1938 (reprint 1996). His English translation (with revisions) is included in *SITI*. III. 2: 1499-1610. The citations are from the English translation unless otherwise stated.

c21.s6.3

³ T.N. Subramanian 1957: Pl. 1, p. 1512.

c21.s6.4

(a) treatment of the consonantal symbol as basic, that is, without the 'inherent' *-a* medial vowel; and
(b) the consequential use of the *-ā* medial stroke of Brāhmī to indicate the *-a* medial vowel,

c21.s6.5

are "found also in the Brāhmī inscriptions of Tamiḻ-nāḍu".⁴ Later discoveries of these orthographic conventions not only in cave inscriptions but also in pottery inscriptions and legends on coins, seals, etc., in the Early Tamil-Brāhmī script, have proved beyond doubt the correctness of the important observation made by Subramanian.

c21.s6.6

⁴ Ibid. pp. 1508-1509.

c21.s6.7

Subramanian also noted that the characteristic *-ā* medial sign of Bhattiprolu does not occur in the Tamil cave inscriptions.⁵ He was, however, uncomfortable, like Subrahmanya Aiyer before him, with the idea that the same sign could represent either *-a* or*-ā* medial vowels according to the context. This explains his search for yet another alternative solution and his claim that the medial **-ā** was sometimes represented by two horizontal strokes attached to the consonantal symbol.¹ The claim is not borne out by actual evidence *in situ*.²

c21.s6.8

⁵ Ibid. p. 1509.

c21.s6.9

¹ T.N. Subramanian 1957: p. 1509 and Pl. 1.

c21.s6.10

² Subramanian does not cite the source of his illustration. Apparently his drawing of *cā* with two horizontal strokes is based on the estampage of an inscription at Kongarpuliyankulam (No. *12* in the present Corpus). The letter is read as *ce*/*cē* by most authorities including T.V. Mahalingam (1967: pp. 224-231) or as *cō* by others.

c21.s6.11

Subramanian does not also appear to have comprehended the true import of the Bhattiprolu conventions as seen from his explanation that they were intended to avoid the confusion between the Brāhmī *anusvāra* and the Tamil *puḷḷi*.³ As we shall see presently, the *puḷḷi* was a later development in Tamil-Brāhmī in response to a problem which did not exist in the Bhattiprolu script.⁴

c21.s6.12

³ T.N. Subramanian 1957: p. 1508.

c21.s6.13

⁴ See sections 6.8 & 6.9.

2.7 The first linguistic study

c21.s7.1

In 1964, Kamil Zvelebil, the renowned Dravidianist, published the first formal study of the language of the cave inscriptions.⁵ The linguistic analysis is based on his readings of 34 inscriptions from 13 sites.⁶ His readings from the photographs of estampages published in *ARE* stay close to those of Krishna Sastri and Subrahmanya Aiyer with only a few relatively minor emendations. It is thus inevitable that Zvelebil's readings share the same problems which have affected the earlier attempts, namely, lack of understanding of the relevant rules of orthography, wrong identification of the earlier forms of ***ḻ*** and ***ṟ***, and the generally incorrect readings based on unsatisfactory estampages. As regards orthography, Subrahmanya Aiyer's rule that the penultimate vowel in a word tended to get unnecessarily lengthened, was extended by Zvelebil to include the "ultimate and penultimate syllables" ⁷ also. However, this modification too does not cover all the observed cases of apparent lengthening of vowels (e.g., **i ḷā nā tā ṉa** for ***iḷanataṉ*** in **No. 17**).⁸ The simpler and more elegant solution to the problem of apparent lengthening of vowels is to recognise the dual values of the medial sign representing **-a** and **-ā** according to the context. In spite of these problems, Zvelebil was able to reach the conclusion that the language of the cave inscriptions is basically Tamil, though in a "hybridized form".⁹ We shall consider the results of Zvelebil's linguistic analysis more fully in the next chapter.¹⁰

c21.s7.2

⁵ Kamil Zvelebil 1964: *Archiv Orientalni*, 32, pp. 647-675 (reprinted in ) *Tamil Culture*, vol. XII, No. 1 (1966): pp. 13-50 (=KZ). Citations are from the latter source.

c21.s7.3

⁶ Excluding Sittannavasal II which belongs to a later period (ca. 8th century A.D.).

c21.s7.4

⁷ Kamil Zvelebil 1966: p. 46.

c21.s7.5

⁸ My earlier attempt to explain the apparent lengthening of the final syllable (-a > -ā) as due to writing the final *-ṉ* as *-aṉ* also failed for the same reason (I. Mahadevan 1965c).

c21.s7.6

⁹ Kamil Zvelebil 1966: p. 48.

c21.s7.7

¹⁰ See section 3.1.

2.8 Discovery of the puḷḷi in the Tamil-Brāhmī script

c21.s8.1

The *puḷḷi* (literally, 'dot') is a diacritic device in the Tamil script to mark the basic consonants and (formerly) the short *e* and *o* vowels.¹ The occurrence of the *puḷḷi* in the cave inscriptions was not suspected for a long time. Krishna Sastri did notice the 'dots' in the Anaimalai cave inscription, but considered them to be the *anusvāra* of the Brāhmī script.² Even Subrahmanya Aiyer failed to notice the occurrence of the *puḷḷi* in this inscription which he discovered. In general, he assumed that "as in most Tamil inscriptions and palm-leaf manuscripts, it (the *puḷḷi*) has to be supplied by the reader".³ This assumption shows that he did not grasp the essence of the Bhattiprolu convention that the consonant symbol does not possess the so-called inherent *-a* medial vowel as pointed out by Bühler much earlier.⁴

c21.s8.2

¹ See sections 5.14, 6.8 and 6.9.

c21.s8.3

² H. Krishna Sastri 1919: p. 334, n.6.

c21.s8.4

³ K.V. Subrahmanya Aiyer 1924: p. 287.

c21.s8.5

⁴ G. Bühler 1894 b: pp. 323-329. See also section 6.6 in this vol.

c21.s8.6

The Arachalur cave inscription discovered in 1961 has a clear dot placed inside the character for the vowel *e*.⁵ The significance of the dot as the *puḷḷi* indicating the short vowel *e* was first pointed out by Krishnan in 1973.⁶ The next find of a *puḷḷi* came from a wholly unexpected source, the Tamil-Brāhmī legend on the reverse of the silver portrait coin of Sātakarṇi (2nd century A.D.). The occurrence of the *puḷḷi* in the coin-legend was first pointed out by Sircar.⁷ This was soon followed by the discovery of the *puḷḷi* in the legend of the Andippatti Hoard of lead coins of Cēntaṉ (ca. 3rd century A.D.) by Mohandas.⁸ The next advance came when Jebarajan and Gift Siromoney identified the *puḷḷi* in the Anaimalai cave inscription, overlooked by the earlier investigators.⁹ It is now known that the earliest occurrence of the *puḷḷi* which can be dated is in the coin-legend on the silver portrait coin of Puḷumāvi issued towards the close of the 1st century A.D.¹⁰

c21.s8.7

⁵ No. **85**.

c21.s8.8

⁶ K.G. Krishnan 1973-74b: pp. 91-93.

c21.s8.9

⁷ D.C. Sircar 1963-64b: pp. 247-252. See also Fig. 5.9 in this vol.

c21.s8.10

⁸ P.N. Mohandas 1968, *The Hindu*, January 7. K.G. Krishnan 1973-74c: pp. 139-140.

c21.s8.11

⁹ E. Jebarajan and Gift Siromoney 1978: pp. 1-10.

c21.s8.12

¹⁰ I.K. Sarma 1980: p. 115; Pl. 9 & 9A. See also section 5.15 in this vol. for discussion on the coin-legends.

2.9 The first book on cave inscriptions

c21.s9.1

In 1967, Mahalingam published the first book-length study of the Tamil cave inscriptions. In this volume, Mahalingam presented his readings and interpretations of 54 inscriptions from 17 sites discovered up to the end of 1962.¹¹ For the first time, almost all the inscriptions included in the volume have been illustrated with photographs of reasonable size supplied by the ASI. About 20 of the photographs appear here for the first time.

c21.s9.2

¹¹ T.V. Mahalingam 1967: pp. 201-298 (=TVM).

c21.s9.3

Publishing his work in the late sixties of the last century, Mahalingam started with several distinct advantages. The language of the cave inscriptions had been proved to be Tamil, and the four special Tamil characters *ḻ*, *ḷ*, *ṟ* and *ṉ* had been identified. The presence of the Bhattiprolu convention and the Tamil *puḷḷi* in the cave inscriptions were known to him. The number of inscriptions had almost doubled providing more material for comparative study.

c21.s9.4

In spite of these advantages, Mahalingam's readings and interpretations are disappointing. The failure is almost certainly due to his reliance on photographs of estampages and his apparent omission to check the readings *in situ*.¹ Had he done so, he could not have missed the longest Pāṇṭiya and Cēra inscriptions respectively at Mangulam and Pugalur not included in his book; ² he would have also avoided the mix-up of estampages at Mangulam, the incorrect order of reading the inscriptions at Kongarpuliyankulam and Mettuppatti ³ and quite a few incorrect readings (like *ṭ* for *ṟ* and *ṭh* for *ḻ*).⁴ Following Subramanian, Mahalingam also claimed to have found the presence of two strokes attached to consonants to indicate the *-ā* medial vowel in two of the inscriptions edited by him.⁵ This claim is also not borne out by actual evidence *in situ*.⁶

c21.s9.5

¹ Except Kunnakkudi and Arachalur caves which he visited (T.V. Mahalingam 1958 and 1962 respectively).

c21.s9.6

² Nos. *1* and *62* respectively.

c21.s9.7

³ T.V. Mahalingam (1967) follows the incorrect order in K.V. Subrahmanya Aiyer 1924.

c21.s9.8

⁴ E.g., *veḷ-aṭai* for *veḷ-aṟai* (T.V. Mahalingam 1967: No. 1 D & E at pp. 210-211); *eṭhuyūr* for *eḻaiy-ūr* (ibid. p. 213).

c21.s9.9

⁵ T.V. Mahalingam 1967: p. 228, n.6; pp. 253-254.

c21.s9.10

⁶ See n.3, No. *11* and n.4, No. *55*.

c21.s9.11

Mahalingam's interpretations, even when the readings are clear, lack plausibility in many cases. For example, he segments *koṭupitavaṉ* (LT *koṭuppittavaṉ* 'he who caused to be given') as *kōṭu* 'hill' and *pitavāṉ*, linking the latter to Skt. *pitā* 'father'; ⁷ he reads *poṉ* 'gold' for which he suggests the alternative meanings of 'Mount Mēru' or 'trap' or 'cavern'; ⁸ he reads *ita* and interprets it as 'Indra'; ⁹ the personal name *ātaṉ* is derived from Skt. *āpta* and interpreted as "a holy person belonging to the Ājīvika faith"; ¹⁰ in none of these cases, the suggested interpretations fit the context.

c21.s9.12

⁷ T.V. Mahalingam 1967: p. 228.

c21.s9.13

⁸ Ibid. pp. 229-230.

c21.s9.14

⁹ Ibid. pp. 230-231.

c21.s9.15

¹⁰ Ibid. p. 287.

2.10 Renewed study of cave inscriptions: the first field expedition (1962-66)

c21.s10.1

The cave inscriptions are engraved on uneven, rough and badly weathered rock without any prior dressing of the surface. Estampages (inked impressions) show up pits, depressions and faults on the rock surface, creating a babel of background 'noise' which almost drowns the 'information' of the inscriptions. Except in cases where the inscriptions happen to be favourably located, the estampages tend to be virtually illegible. It is even more difficult to make out the texts from photographs of such estampages, given the scale of reduction, the quality of the photographs and reproduction in print. The earlier attempts to decipher the cave inscriptions were unsuccessful largely due to reliance on unsatisfactory estampages and photographs. This is also the reason why many important inscriptions including a Pāṇṭiya inscription from Mangulam and a Cēra inscription from Pugalur remained unpublished for long periods of time.

c21.s10.2

When I took up the study of the cave inscriptions in 1961, the Government Epigraphist was good enough to place a set of estampages at my disposal. After struggling with them in vain for a few months, I gave up the attempt and decided to visit the caves personally to study each inscription *in situ*. It took me five working seasons mostly in the winter months of 1962-66 to complete the first round of visits to the caves. Where the inscriptions were at inaccessible heights, ladders or scaffolding constructed from bamboo, casuarina poles or steel slotted angles were made use of to look at the inscriptions from very close quarters and to prepare transcriptions and tracings. Most of the caves were visited more than once to study the inscriptions when sunlight fell on them at different angles during the day.¹

2.11 Tracings of cave inscriptions

c21.s11.1

Inscriptions are normally illustrated with mechanically prepared estampages. Direct photographs of stone inscriptions, except in very favourable circumstances, are seldom satisfactory, and eye copies are not acceptable as they are liable to subjective error. However, an extraordinary situation prevailed in the case of the Tamil cave inscriptions calling for extraordinary remedies. Estampages had been published in the *ARE* and experts of the calibre of Krishna Sastri and Subrahmanya Aiyer had attempted to read them but not too successfully. There was a stalemate with no further progress for four decades between 1924 and 1964. Photographs of fresh estampages made by the ASI in 1963-64 were not noticeably better than the earlier ones. The option of publishing my readings without illustrations did not appeal to me. I was eager to communicate to the scholarly world what I saw–and what I saw was quite different from what had been published till then. It was in these exceptional circumstances that I adopted the unconventional course of preparing tracings direct from the stone and publishing them in support of my revised readings of the cave inscriptions.

c21.s11.2

The following is a brief account of how the tracings were made (Fig. 2.¹). The rock surface was first cleaned to remove dirt, moss, etc. The grooves of the inscriptions were marked with chalk for better visibility through the translucent tracing paper which was then affixed to the rock surface with adhesive tape. The paper was then carefully squeezed into the grooves and a pencil was run along the sharp edges of the grooves to delineate the outline of the letters. After some experimentation and practice, we were able to make reasonably accurate tracings. Generally, more than one tracing was taken of each inscription. Finally, fair copies of the tracings were made in Indian ink for publication.

c21.s11.3

¹ R. Nagaswamy, Director of Archaeology, Tamil Nadu, provided all the necessary facilities for the field work and accompanied me to many of the caves. R. Selvakumaraswamy, Executive Engineer, provided technical support especially for erecting the scaffolding. Madagadi K. Thangavelu, civil engineer and expert draughtsman, prepared the tracings under my supervision.

2.12 Corpus of the Tamil-Brāhmī Inscriptions 1966 (1968)

c21.s12.1

A corpus of the cave inscriptions discovered up to the time comprising 74 Tamil-Brāhmī and 2 Early Vaṭṭeḻuttu inscriptions from 21 sites was compiled by me in 1966 based on the following principles.

c21.s12.2

(1) Each inscription was read and edited direct from the stone.¹
(2) The inscriptions were arranged chronologically based on palaeographic, orthographic and linguistic evidence as follows.
| | | |

c21.s12.3

¹ Except IM No. 68/1966 at Tiruchirapalli-A (No. **78** in the present Corpus) which is no longer extant.

c21.s12.4

| Early Period | : | 2-1 centuries B.C. |
| Middle Period | : | 1-2 centuries A.D. |
| Late Period | : | 3-4 centuries A.D. |
| Early Vaṭṭeḻuttu Period | : | 5-6 centuries A.D. |

c21.s12.5

(3) The following rules of orthography for the notation of medial vowels were formulated and rigorously applied to the reading and interpretation of the texts.
> (a) *Earlier period*
>> (i) A consonantal symbol unaccompanied by a medial vowel sign represents the basic consonant only.
>> (ii) The medial *-a* and *-ā* vowels are indicated by the same sign and can be distinguished only from the linguistic context.
>> (iii) The special *-ā* medial sign of the Bhattiprolu script is not found in the cave inscriptions.
> (b) *Later period*
>> (i) A consonantal symbol is treated either as basic or with the 'inherent' medial *-a* depending on the linguistic context.
>> (ii) The *puḷḷi* does not occur in the cave inscriptions.²
>> (iii) The top horizontal stroke attached to a consonant represents the medial *-ā* only.

c21.s12.6

² I had earlier failed to notice the occurrence of the *puḷḷi* in the Anaimalai and Arachalur cave inscriptions. The orthographic scheme was modified later to accommodate the *puḷḷi*.

c21.s12.7

The *Corpus* was presented (as a pre-printed monograph) at the Seminar on Inscriptions held in Madras in 1966 and included in the Proceedings of the Seminar published in 1968.³ The paper is illustrated with tracings of 46 selected inscriptions. A companion paper explaining the results of the revised readings in the *Corpus* was presented at the Second International Conference Seminar of Tamil Studies held in Madras in January 1968.¹ The revised readings have been widely accepted and form the basis for much of the later work in the field.

c21.s12.8

³ I. Mahadevan, *Corpus of the Tamil-Brāhmī Inscriptions 1966*; (also in) R. Nagaswamy 1968 (ed.) *Seminar on inscriptions 1966*: pp. 57-73 (=IM 1966).

c21.s12.9

¹ I. Mahadevan 1971: pp. 73-106. The discussion on the paper (pp. 104-106) was centred on the finding that "Tamil became a written language for the first time in about 2nd cent. B.C. by the adaptation of the Brāhmī script to the Tamil phonetic system" (p. 94).

2.13 Later publications on Tamil-Brāhmī inscriptions

c21.s13.1

2.13.1 R. Nagaswamy 1972

c21.s13.2

This is a Tamil version of the texts of the Tamil-Brāhmī inscriptions (included in the *Corpus 1966*) with brief comments suggesting modifications in some of the readings and interpretations.² In particular, his reading and interpretation of the Arachalur musical inscriptions have superseded the earlier unsuccessful attempts by Mahalingam and myself.³ Another major improvement is the correction of my reading *mātavirai* 'great (Buddhist) nun' to *mātatirai* (> *matirai*) 'the city of Madurai', in an inscription from Alagarmalai, restoring the earlier reading by Krishna Sastri and interpretations by Subrahmanya Aiyer and Mahalingam.⁴ This is an important correction as it has done away with the only substantial evidence for associating the cave with the Buddhist faith. Nagaswamy also agreed with Mahalingam in reading *ivakuṉṟa-* 'elephant hill' in the Anaimalai cave inscription.⁵ He drew my attention to the occurrence of the *puḷḷi* in the stone-bed inscriptions at Pugalur.⁶ Another significant improvement due to Nagaswamy is the revised reading *perunta(c)caṉ* 'master mason' in the Early Vaṭṭeḻuttu inscription at Pillaiyarpatti, confirming that the excavation of the rock-cut shrine and the engraving of the inscription were coeval.⁷ This has helped in dating Pillaiyarpatti, the earliest of the rock-cut shrines in the Pāṇṭiya country, in the 6th century A.D. on palaeographic evidence. I have incorporated these and a few other minor improvements suggested by Nagaswamy in the present *Corpus*.

c21.s13.3

² R. Nagaswamy 1972d: pp. 48-80 (=RN). He has dealt with 72 out of 76 inscriptions included in the *Corpus 1966*.

c21.s13.4

³ Ibid. pp. 75-79.

c21.s13.5

⁴ Ibid. p. 60.

c21.s13.6

⁵ Ibid. pp. 62-63.

c21.s13.7

⁶ Ibid. p. 68.

c21.s13.8

⁷ Ibid. pp. 79-80. See also R. Nagaswamy 1973: p. 69.

c21.s13.9

2.13.2 R. Panneerselvam 1972

c21.s13.10

This is a critical study of 46 inscriptions selected from the *Corpus 1966*.⁸ The tracings which illustrate these inscriptions in the *Corpus* have been reproduced in his paper. While closely following the readings and interpretations in the *Corpus*, Panneerselvam has also suggested a few modifications. Among the improvements made by him and incorporated by me in the present *Corpus* are the revised segmentation of the longer Mangulam inscriptions¹ and the Sittannavasal inscription² leading to better understanding. Panneerselvam has also included in his paper a detailed linguistic study of the inscriptions. A comparison of Panneerselvam's inscriptional glossary and linguistic analysis with those by Kamil Zvelebil (1964) provides a measure of the progress achieved in understanding the language of the cave inscriptions.

c21.s13.11

⁸ R. Panneerselvam 1972: pp. 163-197 (=RPS).

c21.s13.12

¹ R. Panneerselvam 1972: Nos. 1 to 3 at p. 183.

c21.s13.13

² Ibid. No. 22 at p. 185.

c21.s13.14

Panneerselvam has also made a most valuable contribution by his identification of three generations of the Irumporai princes mentioned in the Pugalur inscriptions with the heroes of the Seventh to Ninth decades of *Patiṟṟuppattu*, establishing for the first time a definite link between the kings celebrated in Caṅkam literature and those mentioned in the Tamil-Brāhmī inscriptions.³ Panneerselvam's identification has won acclaim.

c21.s13.15

³ R. Panneerselvam 1968: pp. 421-425.

c21.s13.16

2.13.3 Mayilai Seeni Venkataswamy 1981

c21.s13.17

This study of the Tamil-Brāhmī inscriptions published posthumously in 1981 is the second book-length treatment of the subject, and the first to appear in Tamil.⁴ This volume contains readings and interpretations of 46 cave inscriptions from 17 sites. Most of the texts are illustrated with tracings reproduced from the *Corpus 1966*; a few other illustrations seem to be drawings apparently based on the photographs of estampages published by Mahalingam (1967).

First 200 of 365 passages. Read on

Linked texts (4)