Pramāṇa · Search · About · API

Early Tamil Epigraphy: From the Earliest Times to the Sixth Century A.D. c25: 5. Palaeography

Early Tamil Epigraphy: From the Earliest Times to the Sixth Century A.D. c25

Language
English

Contents

5. Palaeography

5.1 Nomenclature

c25.s1.1

Variants of the Brāhmī script are referred to in this study with the following connotations:

c25.s1.2

**Brāhmī**: The general term referring to the earliest known stage of the script from which all other native Indian scripts (except the Harappan) are derived. The term Asokan Brāhmī is used when referring specifically to the script of the edicts; the term Mauryan Brāhmī has a somewhat wider significance and includes other inscriptions assigned to the Mauryan Age.¹

c25.s1.3

¹ An enquiry into the origin of the Brāhmī script is not within the scope of the present study. For a good introduction to the subject, see S.P. Gupta and K.S. Ramachandran (eds.) 1979.

c25.s1.4

**Northern Brāhmī**: Post-Mauryan Brāhmī of Northern India including north-western, central and north-eastern regions.

c25.s1.5

**Southern Brāhmī**: Post-Mauryan Brāhmī of the Deccan including Western Deccan. The term 'Southern Brāhmī', as employed in this study, does not include the Bhattiprolu and Tamil-Brāhmī scripts.

c25.s1.6

**Bhattiprolu script**: The unique variant of Brāhmī found only in the casket inscriptions at this site in Andhra Pradesh.

c25.s1.7

**Sinhala-Brāhmī**: The variant of Brāhmī adapted to the Sinhala-Prakrit language of the cave inscriptions in Sri Lanka. The Sinhala-Brāhmī script has two rather sharply differentiated phases viz., Early Sinhala-Brāhmī (ca. 2-1 centuries B.C.) and Late Sinhala-Brāhmī (ca. 1-7 centuries A.D.).²

c25.s1.8

² The dates are based on S. Karunaratne 1984. The recent claims for a much earlier date for pottery inscriptions in Brāhmī found at Anuradhapura, Sri Lanka, are not considered in the present study. See S.U. Deraniyagala 1992: pp. 739-750; F.R. Allchin 1995: pp. 176-179; R.A.E. Coningham *et al.* 1996: pp. 73-97.

c25.s1.9

**Tamil-Brāhmī**: The script of the earliest known Tamil inscriptions is referred to as Tamil-Brāhmī as it is an adaptation of Brāhmī for writing in Tamil.³ Unlike all other regional variants of Brāhmī, Tamil-Brāhmī is uniquely adapted for a non-Indo-Aryan, Dravidian language. This fact and the consequential palaeographic and orthographic modifications effected in the script entitle it to the status of a separate script.⁴ This has been recognised even in ancient times as attested by the names *dāmili* or *drāviḍī* found in the Jaina canonical texts *Paṇṇavaṇā-* *sutta* and *Samavāyāṅga-sutta*, and the name *drāviḍa-lipi* in the later Buddhist work *Lalitavistara*, which probably refer to different stages of the script.

c25.s1.10

³ The term 'Tamil-Brāhmī' has been criticised by some scholars who object to the prefix 'Tamil-' as it offends against the 'imperial' unity of Brāhmī; and by others who object to the suffix '-Brāhmī' as they deny that the Tamil script is derived from Brāhmī.

c25.s1.11

⁴ E.g., Greek from Phoenician, and Kharoṣṭhī from Aramaic. Scripts also acquire new names when their appearance is radically transformed in the course of evolution; e.g., Brāhmī evolving into Nāgarī and Tamil-Brāhmī into Vaṭṭeluttu. A separate script does not necessarily imply an independent origin.

c25.s1.12

Bühler identified Drāviḍī with the variant of Brāhmī found at Bhattiprolu.¹ However, as the Bhattiprolu inscriptions are in Prakrit, it seems inapt to describe this script as 'Drāviḍī' except perhaps in the geographical sense. Attempts have also been made to revive the names Drāviḍī² and Dāmilī³ to refer to the script of the earliest Tamil inscriptions. However, the descriptive term Tamil-Brāhmī, which transparently indicates Tamil as the language and Brāhmī as the script, has gained wide acceptance. The evolution of the Tamil-Brāhmī script is divided into two broad periods viz., Early Tamil-Brāhmī (ca. 2nd century B.C.-1st century A.D.) and Late Tamil-Brāhmī (ca. 2nd-4th centuries A.D.).⁴

c25.s1.13

¹ G. Bühler 1896 (1959 reprint): p. 31.

c25.s1.14

² T.V. Mahalingam 1967: pp. 201-298. He introduces each inscription read by him as 'in Drāviḍī'.

c25.s1.15

³ R. Nagaswamy 1972a: pp. 9-10. He has coined the name Tamiḻi (from Pkt. *dāmilī* ).

c25.s1.16

⁴ For chronology of Tamil-Brāhmī, see discussion in section 2.16.

c25.s1.17

**Early Vaṭṭeḻuttu:** Late Tamil-Brāhmī script evolved into Early Vaṭṭeḻuttu (ca. 5th-6th centuries A.D.). The later evolution of Vaṭṭeḻuttu is outside the scope of the present study.

c25.s1.18

**Early Tamil inscriptions:** Tamil inscriptions in Tamil-Brāhmī and Early Vaṭṭeḻuttu up to the end of the 6th century A.D. are referred to as Early Tamil inscriptions in this study.

c25.s1.19

**Tamil script** which begins to appear in the Pallava territory from the 7th century A.D. falls outside the chronological limits of the present study.⁵

c25.s1.20

⁵ However, the palaeographic developments in Vaṭṭeḻuttu and Tamil scripts from the 7th century A.D. are briefly surveyed in the concluding sections 5.19 & 5.20 in this chapter.

5.2 Palaeographic Charts

c25.s2.1

The discussion on palaeography is illustrated by a set of eight palaeographic charts appended to this chapter. Their format is described below:

c25.s2.2

Chart 1. **The Brāhmī script:** The chart illustrates the earliest known forms of the letters of the Brāhmī script. These are based on the neat monumental forms of the Asokan Pillar Edicts; the few letters not found in the edicts are taken from the earliest available sources.⁶

c25.s2.3

⁶ The primary source for Asokan Brāhmī is E. Hultzsch, *CII*. I (1991 reprint). I have also consulted C.S. Upasak 1960. For the post-Asokan palaeography, the following sources have been consulted: G. Bühler 1896 (1959 reprint); G.H. Ojha 1918 (1993 reprint); A.H. Dani 1963 (Indian edition 1986) and C. Sivaramamurti 1952.

c25.s2.4

Chart 2. **The Tamil-Brāhmī script:** The chart presents in a normalised form the earliest known occurrences of the letters of the Tamil-Brāhmī script as found in the cave inscriptions.

c25.s2.5

Chart 3. **Origin and evolution of additional letters in Tamil-Brāhmī:** The chart¹ illustrates the origin of the four additional letters viz., *ḻ*, *ḷ*, *ṟ* and *ṉ* in Tamil-Brāhmī. The forms are normalised to show the mode of formation from Brāhmī and the main stages in the evolution of the letters.

c25.s2.6

¹ This is a revised and updated version of the chart published by me earlier (I. Mahadevan 1971: p. 102).

c25.s2.7

Chart 4. **Evolution of vowels**

c25.s2.8

Charts 5A & B. **Evolution of consonants**

c25.s2.9

Chart 6. **Evolution of medial vowel signs**

c25.s2.10

Charts 4 to 6 are each divided into three periods viz., Early Tamil-Brāhmī (ca. 2nd century B.C.-1st century A.D.), Late Tamil-Brāhmī (ca. 2nd-4th centuries A.D.) and Early Vaṭṭeḻuttu (ca. 5th and 6th centuries A.D.). The letters included in Charts 4 to 6 have been assembled by computer from direct tracings of the inscriptions in the Corpus. The numbers in italics printed below the letters refer to the inscriptions where the forms occur. Illustrations of the inscriptions may be referred to in case of forms not included in the charts.

c25.s2.11

Chart 7. **Tamil-Brāhmī: special forms from pottery inscriptions:** Special forms of the Tamil-Brāhmī letters found in the pottery inscriptions, which occur rarely or not at all in the stone inscriptions, are illustrated in this chart from the following sources:

c25.s2.12

Arikamedu: R.E.M. Wheeler *et al.* 1946: Figs. 46 & 47; Pl. XLI. (abbr. AKW). Vimala Begley *et al.* 1996: Figs. in chapter 5. (abbr. AKB).

c25.s2.13

Kodumanal: Y. Subbarayalu, Catalogue of Pottery Inscriptions from Kodumanal 1996 (unpublished). (abbr. KDM).

c25.s2.14

Other Sites: Natana Kasinathan, *JESI* 23 (1997): Pl. II-VII. (Karur: abbr. KRR; Alagankulam: abbr. ALA).

c25.s2.15

Chart 8. **Early Vaṭṭeḻuttu script at Pulankurichi:** The chart² illustrates forms from the Pulankurichi inscriptions which are not included in the present Corpus. It is, however, necessary to take note of the forms occurring in them to trace the course of evolution from Late Tamil-Brāhmī to Early Vaṭṭeḻuttu.

c25.s2.16

² The chart is based on eye copies of the inscriptions made by S. Rajagopal and charts of the letters published in V. Vedachalam 2001: pp. 19-26. I have also studied the inscriptions *in situ*.

5.3 Origin of writing in Tamil: earlier theories

c25.s3.1

According to tradition, the Tamil language was received by Agastya from Lord Śiva himself.³ There are, however, no traditional accounts in Tamil regarding the origin of writing as such.

c25.s3.2

³ Cf. e.g., *kaṭavuḷ tanta tamiḻ tantāṉ* 'He (Akattiyan) gave Tamil which he received from the Lord (Śiva)' (*Kampa. Akattiya.* 41). According to the Tamil Buddhist tradition, Agastya received the Tamil language from Avalōkitēśvara (*Vīracōḻiyam, Pāyiram*, 2).

c25.s3.3

*Tolkāppiyam*, the earliest extant grammar in Tamil, refers to the alphabet of thirty letters *a* to *ṉa* (*Tol. Eḻu*.1) and the *puḷḷi* (ibid. 15-16), indicating that the inner character of the writing at that time was the same as in the modern Tamil script. *Naṉṉūl*, the Tamil grammar (ca. 12th century *A.D.*) next only to *Tolkāppiyam* in importance, declares that "all letters are of ancient form" (*Naṉṉūl* 97). While linguistic changes with the passage of time are noticed in the grammatical treatises and the commentaries,¹ the original shapes of letters or their evolution with time are not referred to.

c25.s3.4

¹ Cf. *e.g.*, *paḻaiyaṉa kaḻitalum putiyaṉa pukutalum ... kāla vakaiyiṉāṉē* 'Old (usages) become obsolete and new ones enter (the language) by efflux of time' (*Naṉṉ*. 461).

c25.s3.5

There are some allusions to different types of letters (*eḻuttu*) in medieval commentaries and *nikaṇṭus*. Subramanian who has compiled the references and commented exhaustively on them, identifies the technical expressions occurring in them with successive stages of the evolution of writing from the pictographic to the syllabic and alphabetic.² However, considering the lateness of these literary sources (ca. 8-11 centuries *A.D.*), they are more likely to refer to symbols of various kinds known to have been employed in the medieval inscriptions and palm-leaf manuscripts.

c25.s3.6

² T. N. Subramanian 1957: pp. 1576-1586.

c25.s3.7

Some Tamil scholars deny that the Tamil scripts are derived from Brāhmī. According to them, the high antiquity of the Tamil language and culture presupposes the existence of writing from very ancient times. I shall cite only one example of this view from the authoritative and highly influential *History of Tamil Literature* by Varadarajan:³

c25.s3.8

³ Mu. Varadarajan 1972, *Tamiḻ ilakkiya varalāṟu* (12th edn. 1999): p. 6.

c25.s3.9

Some regard incorrectly that the Vaṭṭeḻuttu and the present-day Tamil script are evolved from Brāhmī. The Brāhmī script current in South India, which is different from the North Indian Brāhmī, is known as Southern Brāhmī as it has developed under the influence of Vaṭṭeḻuttu. Even before the origin and spread of Brāhmī, the Tamils had developed a script of their own and employed it for commerce, literature and other purposes.

c25.s3.10

The claim has a curious parallel in similar claims based on the high antiquity of Vedic literature for the existence of writing in Sanskrit from the earliest times.⁴ Such claims fail to make a clear distinction between the language and the script. All languages of the world are far older than the scripts devised for recording them.

c25.s3.11

⁴ Cf. *e.g.*, Raj Bali Pandey 1952: p. 21: "Thus the traditions of the country, the testimony of foreign writers, literary evidences and positive palaeographic survivals all tend to prove a very high antiquity of the art of writing in India, stretching in the past upto the fourth millennium *B.C.*"

c25.s3.12

Subramanian accepts that the Tamil scripts are derived from the Mauryan Brāhmī script, but contends that the latter is, in its turn, derived from a still earlier script devised for a Dravidian language:

c25.s3.13

The Brāhmī script was originally devised for the Dravidian language and later on adopted for Prākrit which was evolved as a common language and which was more close to the Dravidian. Later on when Sanskrit elements were introduced into the Prākrit and it became more Sanskritised,

c25.s3.14

additional symbols had to be introduced to suit the needs of the new letters and in the course of this process, some symbols not required for the Sanskritised Prākrit got eliminated. And finally when it became the official language it spread throughout the length and breadth of the country with royal authority behind, and displaced the then existing varieties of the script. It should be noted side by side here that no inscription in the other coexisting scripts has been found so far. It is from this Brāhmī script that all the existing scripts of the country including the Dēvanāgarī have been obtained by evolution.¹

c25.s3.15

¹ T.N. Subramanian 1957: pp. 1608-1609.

c25.s3.16

As Subramanian himself admits candidly, there is no inscriptional evidence to support his claim. Further, his theory that the Prakrit languages were originally 'Dravidian' and later became 'Sanskritised' is not generally accepted. Modern linguistic research in Indo-Aryan and Dravidian has established that the Middle Indo-Aryan (Prakrit) languages are descended from the Old Indo-Aryan (Vedic), but are influenced by substratum languages including Dravidian.²

c25.s3.17

² For discussion on the substratum influence of Dravidian on Indo-Aryan, see Asko Parpola 1994: pp. 167-169.

c25.s3.18

Krishnan has suggested a different scenario for the origin of the Tamil script. According to him:

c25.s3.19

The Tamil Script has a long history almost parallel beginning with the Brāhmī script of the times of Asoka. The script is found originally designed to write inscriptions all over India allowing for modifications in respect of Tamil Nadu and for additions in respect of other areas. The modifications in respect of Tamil Nadu are close to the primary letters of the alphabet. The processes involved in the evolution are along known lines such as the desire for change and economy.³

c25.s3.20

³ K.G. Krishnan 1981c: p. 97.

c25.s3.21

He amplifies this idea further in another paper:

c25.s3.22

In the early times in the history of writing, the Brāhmī script which was current in India had a group of letters consisting of nine vowels (*a*, *ā*, *i*, *ī*, *u*, *ū*, *ē*, *ai* and *ō*) and fourteen consonants (*k*, *ṅ*, *c*, *ñ*, *ṭ*, *ṇ*, *t*, *n*, *p*, *m*, *y*, *r*, *l* and *v*) common to all the areas irrespective of the language for which they were used. The additional letters such as the 15 aspirated letters and 4 letters (*s*, *ś* *sh* and *h*) in the case of the Northern alphabets, and 4 letters (*ḻ*, *ḷ*, *ṟ* and *ṉ*) in the case of the Dravidian alphabets have been specially devised to serve the respective languages that used those alphabets.⁴

c25.s3.23

⁴ K.G. Krishnan 1974: pp. 29-30.

c25.s3.24

Apart from the fact that there is no evidence for such a 'primary national script' before the time of Asoka, it seems most unlikely that a script could have been devised without reference to the language it records. There is no such instance in the history of the writing systems of the world before the invention of the International Phonetic Alphabet (IPA) for modern linguistic research.

c25.s3.25

In recent years there have been many attempts to derive the Brāhmī/Tamil-Brāhmī scripts from megalithic and chalcolithic graffiti, and through them, ultimately from the Indus script.⁵ In so far as such research relates to the decipherment of the Indus script on the basis of comparisons with the pottery graffiti, it falls outside the scope of the present study.¹ However, the Tamil-Brāhmī script is unlikely to have been derived from the megalithic graffiti as stratigraphic evidence from the excavations, especially at Uraiyur² and Kodumanal,³ show the coexistence of the two systems (sometimes on the same pottery as at the latter site), though the origin of the graffiti goes back to a much earlier period.

c25.s3.26

⁵ A.V. Narasimha Murthy 1992: pp. 57-66. R. Madhivanan 1995: pp. 49-79. S. Gurumurthy 1999: pp. 149-163.

c25.s3.27

¹ For my views on the subject, see I. Mahadevan 1995a: pp. 2-8.

c25.s3.28

² P. Shanmugam 1983: pp. 31-36.

c25.s3.29

³ K. Rajan 1994: pp. 76-83.

c25.s3.30

Just as there have been attempts to prove the Brāhmī script to be an artificial creation not derived from any pre-existing source,⁴ there have also been similar attempts in respect of the Tamil script. In 1917, Manickam Naicker, an engineer by profession and learned Tamil scholar, published a monograph⁵ which attempted to prove with the help of detailed diagrams that the letters of the Tamil script and the mystic symbol *aum* (*ōm*) can be derived by cutting and pasting segments of the first letter *a*. The fact that he employs the modern forms of the Tamil letters and of the symbol for *ōm* does not seem to have bothered him.

c25.s3.31

⁴ E.g., A. Cunningham 1877: *CII.* I: pp. 53 ff. He follows the acrophonic principle (e.g., *ka* from *kartari* 'dagger', *dha* from *dhanu* 'bow', *sa* from *sarpa* 'snake', etc.).

c25.s3.32

⁵ P.V. Manickam Naicker 1917: pp. 1-17.

c25.s3.33

Gift Siromoney and Michael Lockwood⁶ claim that "the Brāhmī script was invented at one stroke—possibly by one individual"; they reject "both the theory that it was evolved from the Indus script and also the theory that it was borrowed and developed from some non-Indian script". The point of departure for their demonstration is the pair of symbols found at the end of two Tamil-Brāhmī inscriptions at Kongarpuliyankulam.⁷ They proceed to normalise the symbols into "basic geometric patterns" consisting of a square, a circle, a cross and a vertical line. From segments of these components, they "extract" all the letters of the Brāhmī script.⁸ As Gelb points out, signs of the linear scripts are likely to show some resemblance to one another as these are based in practice on a few simple forms of straight lines, triangles, squares and circles which can be easily remembered.⁹

c25.s3.34

⁶ Gift Siromoney and Michael Lockwood 1992: pp. 219-223.

c25.s3.35

⁷ Corpus: Nos. *12* & *13*. See also section 5.18 and Fig. 5.11 (A & B).

c25.s3.36

⁸ Perhaps the authors of this paper meant it to be a spoof on far out theories.

c25.s3.37

⁹ I. J. Gelb 1963: p. 144.

c25.s3.38

The literary evidence in Tamil with datable links to the outside world includes the presence of the Jaina and Buddhist creeds in the Tamil country, references to the Nandas and the Mauryas and graphic descriptions of the maritime trade with the *yavanas* (Romans). This evidence does not stretch beyond *ca.* 3rd century B.C. The archaeological data from pottery inscriptions in the Tamil-Brāhmī script also indicate the same date as the upper limit for writing. There is no material evidence for the existence of any indigenous or independent script for Tamil earlier to or contemporaneous with the Tamil-Brāhmī script.¹ Whatever writing has been discovered on inscribed pottery from the excavated sites in Tamil Nadu, is exclusively in the Tamil-Brāhmī script and occurs only from the megalithic and Early Historical levels dated not earlier than the end of the 3rd century B.C.²

c25.s3.39

¹ This discussion does not include the Indus script, which has not yet been deciphered, nor the megalithic graffiti. In the present state of our knowledge, it seems best to assume that the megalithic graffiti do not constitute a script with phonetic values, but may be mnemoic or representational devices derived from pictorial art and belonging to the category of forerunners of writing (I.J. Gelb 1963: pp. 24-51). For discussion on the symbols occurring with the inscriptions in the Corpus, see section 5.18.

c25.s3.40

² See section 1.13.

5.4 Origin of Tamil-Brāhmī

c25.s4.1

Tamil-Brāhmī is derived from Brāhmī. Compare Palaeographic Charts 1 and 2 showing the standard forms of Brāhmī and Tamil-Brāhmī scripts respectively. The evidence, which is quite conclusive, may be briefly summarised as follows:

c25.s4.2

(a) All but 4 of the 26 letters in Tamil-Brāhmī (i.e., more than four-fifths of the total) are identical or nearly so with the corresponding letters of Brāhmī and have the same phonetic values.

c25.s4.3

(b) Even the additional letters in Tamil-Brāhmī viz., *ḻ*, *ḷ*, *ṟ* and *ṉ*, are adapted from letters with the nearest phonetic values in Brāhmī (Palaeographic Chart 3).³

c25.s4.4

³ See section 5.12.

c25.s4.5

(c) The medial vowel signs of Tamil-Brāhmī are also identical with the respective signs in Brāhmī and have the same phonetic values.

c25.s4.6

(d) The alphabetical order of letters common to both the scripts is identical.

c25.s4.7

(e) It is revealing that *Tolkāppiyam* places *ṟ*, *ṉ*, *l* and *ḷ* at the end of the series of stops, nasals and liquids (*Tol. Eḻu.* 19-21). This arrangement deviates from the order based on articulatory phonetics. This small but significant detail indicates that the four special letters were originally regarded as additions to the alphabet taken over from Brāhmī. The additional letter *ṉ* is also described as the last in the Tamil alphabet (ibid. 1).

c25.s4.8

Tamil-Brāhmī is formed by adapting Brāhmī to the requirements of the Tamil phonetic system in the following manner.

c25.s4.9

*Palaeographic changes*

c25.s4.10

(a) Omission of letters for sounds not present in Tamil viz., voiced consonants, aspirates, sibilants, the *anusvāra* (*ṁ*) and the *visarga* (*ḥ*);

c25.s4.11

(b) Addition of letters to represent sounds in Tamil which are not available in Brāhmī viz., *ḻ*, *ḷ*, *ṟ* and *ṉ*;

c25.s4.12

(c) Modification of letters by the employment of a special diacritic mark viz., the *puḷḷi*, to:

c25.s4.13

Figure 5.1. Minor Rock Edict of Asoka at Brahmagiri. Karnataka.

c25.s4.14

Figure 5.2. An inscribed casket from Bhattiprolu. Andhra Pradesh.

c25.s4.15

(i) depict basic consonants in final position;
(ii) avoid ligaturing of consonant clusters; and
(iii) distinguish the short vowels *e* and *o* from the respective long vowels.

c25.s4.16

*Orthographic changes*

c25.s4.17

Along with the above palaeographic changes, several orthographic modifications especially in the notation of medial vowels were also effected in Tamil-Brāhmī. These are discussed in detail in Chapter 6.

5.5 Graphemic inventory of Tamil-Brāhmī

c25.s5.1

There are 26 letters comprising 8 vowels and 18 consonants, and 9 medial vowel signs in Tamil-Brāhmī:

c25.s5.2

Vowels: *a* *ā* *i* *ī* *u* *ū* *e*/*ē* *o*/*ō*
Consonants: *k* *ṅ* *c* *ñ* *ṭ* *ṇ* *t* *n* *p* *m* *y* *r* *l* *v* *ḻ* *ḷ* *ṟ* *ṉ*
Medial vowel signs: *-ā* *-i* *-ī* *-u* *-ū* *-e*/*-ē* *-ai* *-o*/*-ō* and the *puḷḷi*.¹

c25.s5.3

In addition to the above, two letters from the Brāhmī script, viz., *dh* and *s* occur marginally in loanwords from Indo-Aryan in the Early Tamil-Brāhmī inscriptions.²

5.6 Tamil-Brāhmī and other Brāhmī variants

5.6.1 Mauryan Brāhmī

c25.s6.t1.1

An overall palaeographic comparison of the earliest known forms of the common letters between Brāhmī and Tamil-Brāhmī (Charts 1 and 2) leads to the broad judgement that the time of borrowing and adaptation from Brāhmī to Tamil-Brāhmī cannot be later than the end of the Mauryan Age. In general, the Early Tamil-Brāhmī letters share the primitive appearance and the tall and angular forms of the Asokan Brāhmī. (Compare, for example, the Brahmagiri Rock Edict of Asoka in Fig. 5.¹ with the Mangulam grants of Neṭuñceḻiyaṉ in Figs. 8.1 & 8.².) The squarish letters with head-marks ('serifs') of the Northern as well as the Southern Brāhmī scripts of the post-Mauryan era are too evolved to have formed the prototypes for the earliest Tamil-Brāhmī forms.

c25.s6.t1.2

¹ The *puḷḷi* may also be regarded as a 'medial vowel sign' as it functions in effect as a 'minus vowel marker' removing the 'inherent' *-a* from consonant characters and reducing the length of the vowels *ē* and *ō*. See section 5.14 on the palaeographic aspects and sections 6.8 & 6.9 on the orthographic aspects of the *puḷḷi*.

c25.s6.t1.3

² The aspirate *dh* occurs only in the loanword *dhammam*/*dhamam* in two of the earliest inscriptions (Nos. *1 & 2*) at Mangulam. The sibilant *s* occurs 12 times, all in the Early Period, out of which 5 occurrences are also from the inscriptions (Nos. *1-3*) at Mangulam. No other non-Tamil letter occurs in the inscriptions included in the Corpus. Earlier investigators have occasionally misread *ḷ* as *ṭh* (as in Nos. *34 & 49*) or as *ś* (as in No. 7).

5.6.2 Southern Brāhmī

c25.s6.t2.1

There were close cultural, religious and commercial contacts between the Deccan and the Tamil country from the earliest times. However, Southern Brāhmī and Tamil-Brāhmī, both arising from the Mauryan Brāhmī script at about the same time, appear to have evolved independently of each other as shown by the characteristically different forms of letters with common sounds. The Early Tamil-Brāhmī inscriptions (ca. 2nd century B.C.-1st century A.D.) included in the present Corpus do not have any forms influenced by Southern Brāhmī.

c25.s6.t2.2

Influence of Southern Brāhmī is seen for the first time in the Tamil-Brāhmī inscriptions of the Late Period (ca. 2nd-4th centuries A.D.) in the northern and western districts adjoining the Andhra-Karnataka regions. This influence is marginal also in the sense that the few Southern Brāhmī forms seen in these inscriptions do not enter the mainstream of evolution of Tamil-Brāhmī into Vaṭṭeḻuttu. The Southern Brāhmī forms occurring in the Late Tamil-Brāhmī inscriptions are noticed in the following discussion on palaeographic evolution of Tamil-Brāhmī and illustrated in the accompanying charts.

5.6.3 The Bhattiprolu script

c25.s6.t3.1

In 1892 Alexander Rea discovered three relic stone caskets (Fig. 5.2) from a Buddhist stupa at Bhattiprolu, a village in the Krishna District of Andhra Pradesh.¹ Nine Prakrit inscriptions are engraved on the caskets in a unique local variant of the Brāhmī script. Bühler edited the inscriptions in a classic paper elucidating with rare insight their remarkable palaeographic and orthographic 'peculiarities' not known in the Mauryan Brāhmī script.²

c25.s6.t3.2

¹ For the discovery of the Bhattiprolu Caskets, see J. Burgess, 1894: *EI*. II, Preface, pp. ix-xiii.

c25.s6.t3.3

² G. Bühler 1894 b: pp. 323-329.

c25.s6.t3.4

(a) Special palaeographic features

c25.s6.t3.5

Most of the letters of the casket inscriptions resemble those of Mauryan Brāhmī. There are, however, six exceptional forms (*gh*, *c*, *j*, *m*, *l* and *sh*) and one additional letter (*ḷ*) not found in Mauryan Brāhmī. None of these forms is found in Tamil-Brāhmī with the exception of *c* with the tail,³ in which case the resemblance may be merely fortuitous.

c25.s6.t3.6

³ Corpus: Nos. *55*, *105*.

c25.s6.t3.7

(b) Special orthographic features

c25.s6.t3.8

While the special palaeographic features of the Bhattiprolu script have hardly any bearing on the study of Tamil-Brāhmī, the position is entirely different when one considers the special orthographic features of the former script in respect of medial vowel notations. The Bhattiprolu script, which remained as an isolated epigraphical curio until the discovery of the Tamil-Brāhmī inscriptions, has turned out to be the Rosetta Stone for the eventual decipherment of the Tamil-Brāhmī script. As we shall see in Chapter 6, it is the comparison of the orthographic features that provides the decisive evidence that the Bhattiprolu script must be later than the earliest phase of the Tamil-Brāhmī script.⁴

c25.s6.t3.9

⁴ See section 6.10 and Fig. 6.3.

5.6.4 Sinhala-Brāhmī

c25.s6.t4.1

Early Sinhala-Brāhmī has markedly close palaeographic similarities with Mauryan Brāhmī on the one hand, and with Tamil-Brāhmī on the other.¹ Late Sinhala-Brāhmī which is strongly influenced by the contemporary Southern Brāhmī script, most probably due to close contacts with the Buddhist establishments in Andhra, is too divergent from Tamil-Brāhmī of the same period for any useful comparison.²

c25.s6.t4.2

¹ A. Velu Pillai 1980: pp. 38-54; 1981: pp. I-111 to I-120.

c25.s6.t4.3

² S. Paranavitana (1970: p. xxii) notes the similarities between the script of the later cave inscriptions in Sri Lanka and that of Andhra, but attributes the changes to local development.

c25.s6.t4.4

Palaeographic comparison of the earliest forms of the common letters indicates that Sinhala-Brāhmī, like Tamil-Brāhmī, is directly derived from Mauryan Brāhmī before the end of the Mauryan Age.³ In the case of Sinhala-Brāhmī, the chronology is confirmed by the dating of a few royal inscriptions.⁴ Thus the close palaeographic similarities between the common letters of Early Sinhala-Brāhmī and Early Tamil-Brāhmī provide further confirmation for dating the commencement of the latter script also from about the same time (ca. 200 b.c.).

c25.s6.t4.5

³ Cf. palaeographic charts in S. Paranavitana 1970: Pl. facing p. xvi and in S. Karunaratne. 1984: Figs. 1-35.

c25.s6.t4.6

⁴ S. Paranavitana (1970:p. xvii) dates the earliest cave inscriptions in Sri Lanka from the last quarter of the 3rd century b.c. S. Karunaratne (1984:pp. 2-5) places the upper limit of the earliest dated inscriptions at the beginning of the 2nd century b.c. A.H. Dani (1963) has not, even in the revised Indian edition (1986), noticed these works. Dani's dating of the earliest cave inscriptions of Sri Lanka to 'about the beginning of the first century a.d.' (ibid. p. 219) is hardly justified; his dating does not explain wherefrom Sri Lanka could have obtained the Mauryan Brāhmī forms in the 1st century a.d. when they were no longer current.

c25.s6.t4.7

While Tamil-Brāhmī has been adapted to suit the requirements of the Tamil phonetic system, Sinhala-Brāhmī has been adapted to the phonology of Sinhala-Prakrit, which is markedly different from that of the Indian Prakrits. While Tamil-Brāhmī did away with voiced consonants, aspirates, sibilants, the *anusvāra* and the *visarga*, Early Sinhala-Brāhmī dropped long vowels, aspirates, the sibilants *śh* and *s*, the *anusvāra* and the *visarga*, as these sounds did not exist in the language of the early cave inscriptions. While Tamil-Brāhmī added four new letters viz., *l*, *l*, *r* and *n* by modifying the Brāhmī letters with the nearest phonetic values, Early Sinhala-Brāhmī utilised the Brāhmī letter *jh* (an aspirate not present in Early Sinhala-Prakrit) in the place of *j* to represent some allied sound for which no letter was available in Brāhmī.

c25.s6.t4.8

One of the most interesting aspects of Early Sinhala-Brāhmī is the complete absence of ligaturing of consonants. In this respect, Sinhala-Brāhmī is unique among the Prakrit-based variants of Brāhmī. The avoidance of ligaturing of consonants is facilitated in Early Sinhala-Brāhmī by the phonological processes of assimilation and epenthesis which are more thoroughgoing in this language than in Indian Prakrits, most probably due to the substratum influence of Tamil. This would explain how the two scripts, one for a Middle Indo-Aryan and the other for a Dravidian language, were able to avoid ligatures, a prominent feature in all other regional scripts.

c25.s6.t4.9

Other palaeographic features shared by the two scripts are the common forms of the letters **a**, *ī* (read as *i* in Early Sinhala-Brāhmī) *m* and *ḷ*.¹ It is noteworthy that all these forms are replaced in Late Sinhala-Brāhmī by the corresponding forms² derived from Southern Brāhmī. This development indicates that the earlier forms are more likely to be native to the Tamil-Brāhmī script in which they continued to exist and gradually evolved into the corresponding Vaṭṭeḻuttu forms.

c25.s6.t4.10

¹ Compare the forms in Palaeographic Chart 2 with the corresponding forms illustrated in S. Karunaratne 1984: Figs. 1, 2, 26 and 34.

c25.s6.t4.11

² S. Karunaratne, ibid.

5.7 Tamil-Brāhmī: general characteristics of writing

c25.s7.1

Before we take up a detailed palaeographic study of the Tamil-Brāhmī script, we may note some of the general characteristics of writing especially of the cave inscriptions.

c25.s7.2

(1) An important feature of Tamil-Brāhmī is the relatively slower pace of evolution especially in the Early Period when compared with the contemporary Northern and Southern Brāhmī scripts of the post-Mauryan era. This is strikingly illustrated at Arikamedu by the occurrence of a pottery inscription in Hybrid Sanskrit³ in the Northern Brāhmī characters of ca. 1st century A.D. along with pottery inscriptions in Tamil-Brāhmī of about the same date but in which the characters look much earlier when judged in terms of Brāhmī palaeography. The slower pace of evolution is also shared by the Early Sinhala-Brāhmī script of the same period (ca. 2-1 centuries B.C.).⁴

c25.s7.3

³ R.E. M. Wheeler *et al.* 1946: Pl. XLI, No. 3.

c25.s7.4

⁴ See the palaeographic charts (chronologically arranged) in Karunaratne 1984: Figs. 1-35.

c25.s7.5

(2) The direction of evolution of Tamil-Brāhmī is markedly different from those of the Northern and Southern Brāhmī scripts. This has resulted in different shapes for the common letters of the same period. This is strikingly illustrated by the bilingual coin-legends on the silver portrait coins of the Sātavāhanas, with the letters on the obverse in the Southern Brāhmī script looking different from those on the reverse in the Tamil-Brāhmī script.⁵

c25.s7.6

⁵ See section 5.15 for discussion on these coin-legends.

c25.s7.7

(3) Two general trends in the evolution of the letters are:
(a) reduction in height and equalisation of the arms of the letters, and
(b) conversion of angular and rectilinear letters and medial vowel signs into cursive shapes, which ultimately resulted in the emergence of Vaṭṭeḻuttu (lit., 'the rounded script'). This tendency, a peculiarly southern characteristic, may be traced to the practice of writing on palm leaves with iron stylus.

c25.s7.8

(4) The letters of the Tamil-Brāhmī inscriptions of the Early Period, especially those engraved on the brow over the entrance to the caves, are relatively much larger⁶ and more crudely cut than in the later inscriptions, reflecting an early stage of literacy and also lack of experience in engraving on extremely hard and rough rock surfaces.

c25.s7.9

⁶ E.g., the cave inscription at Marukaltalai (No. **14**) in which some of the letters reach almost 40 cm. in height. The scribe apparently used the cubit (*muḻam*) for measuring, which is ' the length from the elbow to the tip of the middle finger' (*TL*).

c25.s7.10

(5) The inscriptions of the Early Period are written in single lines irrespective of their length.¹ The changeover to compact writing in multiple lines of manageable length took place only in the Late Period.²

c25.s7.11

¹ E.g., the long single-line inscription at Alagarmalai comprising several segments (Nos. **38-45**) measuring more than 850 cm. (excluding spaces between the segments and damaged passages).

c25.s7.12

² Compare the long single-line inscriptions (Nos. *1-3*) at Mangulam in the Early Period with the four-line inscriptions (Nos. *61 & 62*) of about the same length at Pugalur in the Late Period.

c25.s7.13

(6) Changes in style due to 'pressure of writing' resulting in smaller, well-formed letters in a flowing manuscript 'hand', begin to appear from ca. 1st century A.D. and become more pronounced in the Late Period.³

c25.s7.14

³ E.g., the neat and uniform engraving of the inscription at Jambai (No. **59**) of ca. 1st century A.D. Some of the inscriptions (e.g., Nos. *67 & 69*) of ca. 3rd century A.D. at Pugalur are in a flowing manuscript 'hand'.

5.8 Direction of writing

c25.s8.1

The general direction of writing in Tamil-Brāhmī and Vaṭṭeḷuttu scripts is from left to right as in the case of Brāhmī and all other scripts derived from it. However, there are exceptional cases of writing from right to left with the characters turned upside down for a remarkable reason which our team discovered by conducting simple experiments *in situ* as described below.⁴

c25.s8.2

⁴ The idea occurred to me during my first visit to the Kilavalavu cave in 1963 when I was sitting under the brow of the cave and happened to look up at the inscription.

c25.s8.3

(a) Two inscriptions, one at Tirupparankunram (No. **53**) and the other at Muttuppatti (No. **56**) are engraved on the front side of raised stone beds. To an observer standing in front of the beds facing the inscriptions, they appear to be engraved from right to left with the letters turned upside down. However, the observer sitting on the beds and looking down would find the inscriptions running in the normal direction from left to right with the letters in proper orientation and would have no difficulty in reading them. It appears that the scribes engraved the inscriptions in this special manner on the instructions of the donors who wished to be remembered by the resident monks in their prayers.

c25.s8.4

(b) An inscription at Tirupparankunram (No. **54**) is engraved in two segments on the rock wall abutting the head-side of two stone beds. To an observer facing the inscriptions, the first segment is seen above the right bed and the second segment above the left bed. He would also notice that the first segment is engraved from right to left with the letters turned upside down, but the second is engraved normally from left to right and the letters in proper orientation. It is, however, clear that, as in the two inscriptions discussed in (a) above, the scribe intended to engrave both the segments of this inscription in such a way that the monks lying on the beds resting their heads on the stone pillows and looking up at the inscription would be able to read it normally. The scribe must have been, however, careless and not remembered to engrave the second segment from right to left and turn the letters upside down.

c25.s8.5

(c) Two more inscriptions, one at Kilavalavu (No. *10*) and the other at Kunnakkudi (No. *74*) are engraved on the brow of the caves. It appears that in these cases also the scribes intended to engrave the inscriptions from right to left with the letters turned upside down so that the monks in the cave looking up at the inscriptions could read them normally. However, the scribes bungled the jobs. At Kilavalavu, the inscription is engraved from right to left, but some of the letters are not turned upside down as required. At Kunnakkudi, the letters are all turned upside down, but the inscription is engraved from left to right!

c25.s8.6

As the experiments performed *in situ* reveal, the Tamil-Brāhmī inscriptions described above are not real exceptions to the normal direction of writing, and there is no reason to believe that the Tamil-Brāhmī script was at any earlier stage written from right to left. Indeed, these exceptions themselves prove the rule, as in three out of the five instances in question, the scribes could not successfully reverse the direction of writing to which they were normally accustomed.¹

c25.s8.7

¹ In these exceptional cases, the illustrations in the present volume are printed from left to right for normal reading. One of the inscriptions (No. *10*) has some letters turned upside down even with respect to the normal orientation.

c25.s8.8

According to Bühler, the Brāhmī script was written in the earliest period "both from the right to the left and from the left to the right".² His theory, based on a reversed coin-legend from Eran, is nowadays not generally accepted especially as coin-legends are notoriously prone to reversal. The Minor Rock Edicts of Asoka at Erragudi,³ which have irregularly engraved lines running in both directions, are very crudely executed and cannot be relied upon to indicate a supposedly earlier trend of right to left writing of Brāhmī.

c25.s8.9

² G. Bühler 1896 (1959 reprint): p. 23.

c25.s8.10

³ D.C. Sircar 1979 (2000 reprint): p. 4 and Pl. II.

c25.s8.11

Sri Lanka has the largest number of Early Brāhmī cave inscriptions written from right to left and with correspondingly reversed (and rarely inverted) orientation of letters. Paranavitana⁴ and Karunaratne⁵ who have published these exceptional inscriptions express themselves in favour of an original right to left direction for Brāhmī. However, there is hardly any evidence from Indian inscriptions in support of this theory. One can only presume that the exceptional right to left direction was a local development peculiar to Sri Lanka in the post-Mauryan period.

c25.s8.12

⁴ S. Paranavitana 1970: p. xxii, n. 3.

c25.s8.13

⁵ S. Karunaratne 1984: pp. 7-8.

5.9 Palaeography of Tamil-Brāhmī and Early Vaṭṭeḻuttu

c25.s9.1

As the palaeography of Brāhmī has been studied in several standard publications,⁶ the present study will concentrate on the special and less familiar aspects of Tamil-Brāhmī and trace the distinct course of its evolution into Early Vaṭṭeḻuttu. The study is based mostly on the stone inscriptions included in the present Corpus. However, the more recently discovered and still not adequately published pottery inscriptions and the Early Vaṭṭeḻuttu inscriptions at Pulankurichi have also been utilised to present a more complete picture of the palaeographic evolution.

c25.s9.2

⁶ See especially G. Bühler 1896, G.H. Ojha 1918, C. Sivaramamurti 1952, C.S. Upasak 1960 and A.H. Dani 1963.

5.10 Vowels

c25.s10.1

**a / ā:** In Early Tamil-Brāhmī, the two arms of the letter are wide apart and often turn away from each other at right angles (Chart 4: *a: 1, 2*). This form is attested in the Asokan Edicts at Erragudi¹ and is also characteristic of the Early Sinhala-Brāhmī script.² The exceptional form (ibid. *63*) looking somewhat like X may be compared with those occurring in the pottery inscriptions at Arikamedu (Chart 7: AKW 6; AKB 7, 20), in which the lower arm of the letter is reversed and placed on the opposite side of the vertical. In Early Vaṭṭeḻuttu, the Late Tamil-Brāhmī forms are still retained at Pulankurichi (Chart 8), but later become cursive (Chart 4: *a: 108, ā: 116*).

c25.s10.2

¹ C.S. Upasak 1960: Appendix No. 1 (9).

c25.s10.3

² S. Paranavitana 1970: chart facing p. xvi.

c25.s10.4

**i / ī:** The letter *i* consists of three dots (Chart 4: *i: 89, 76*) or three short, horizontal dashes (ibid. *17, 60*) both arranged in a triangular pattern. Even though the three-dot form occurs later than the three-dash form in the Corpus, the former is the original form as it is attested in Asokan Brāhmī.³ In Early Vaṭṭeḻuttu, the three dashes join together in the composite form seen at Pulankurichi (Chart 8). At the next stage, this form is simplified as a single, smooth curve (Chart 4: *i: 111*).

c25.s10.5

³ C.S. Upasak 1960: Appendix No. 3.

c25.s10.6

The letter *ī* does not occur in Asokan Brāhmī. The Northern Brāhmī form of four dots in a rectangular pattern (Chart 1), derived from the three-dot form of *i*, does not occur in Tamil-Brāhmī. In this script, there is a special form for *ī* consisting of a tall vertical line flanked by a pair of dots (Chart 2). The letter occurs from the earliest times in the Tamil-Brāhmī inscriptions (Chart 4: *ī: 1*) and its form remains stable. It is possible that Tamil-Brāhmī was formed at a time when Brāhmī still did not have a character for *ī* and one had to be created locally.

c25.s10.7

It appears that there was also a practical reason why Tamil-Brāhmī did not adopt the four-dot form of *ī*, and seldom employed even the three- dot form of *i* which is in most cases represented by the symbol for *ī* (to be read as *i* from the context).⁴ The dots would be virtually invisible on the rough, uneven and undressed rock surfaces on which the Tamil-Brāhmī inscriptions are engraved.

c25.s10.8

⁴ The use of *ī* to denote the sound *i* in the Early Tamil-Brāhmī inscriptions is discussed in section 6.14.

c25.s10.9

The Early Sinhala-Brāhmī cave inscriptions also adopted the more visible Tamil-Brāhmī letter for *ī* for apparently the same practical reason but to denote the sound *i*.⁵ The change in value caused no confusion in Sinhala-Brāhmī as the Sinhala-Prakrit of the early cave inscriptions had no long vowels.⁶ In Late Sinhala-Brāhmī, the Tamil-Brāhmī form was replaced by the contemporary Southern Brāhmī form for *i*.⁷ There is, however, no doubt that the original value of the letter is *ī* as in the Brāhmī inscriptions of India.⁸ Though there are no examples of *ī* in the Late Tamil-Brāhmī or Early Vaṭṭeḻuttu inscriptions included in the Corpus, the letter must have reverted fully to its original long value sometime after the Early Tamil-Brāhmī Period, as the Vaṭṭeḻuttu and Grantha forms ¹ and even the Sinhalese form² of *ī* are all clearly derived from it.

c25.s10.10

⁵ S. Paranavitana 1970: p. xviii.

c25.s10.11

⁶ Ibid.

c25.s10.12

⁷ S. Karunaratne 1984: p. 16; Fig. 2.

c25.s10.13

⁸ C. Sivaramamurti 1952: chart for *ī* (Fig. 23, p. 64).

c25.s10.14

¹ C. Sivaramamurti 1952: chart for *ī* (Fig. 23, p. 64).

c25.s10.15

² S. Karunaratne 1984: Fig. 2.

c25.s10.16

***u* / *ū*:** The letter *ū* does not occur in Asokan Brāhmī presumably on account of its rarity in the language of the edicts. However, the letter is attested in its primitive form in Tamil-Brāhmī ³ from the earliest times (Chart 4: *ū*:*9*). The forms of *u* and *ū* remain stable in Tamil-Brāhmī and Early Vaṭṭeḻuttu. The employment of the letter *ū* for the sound *u* in the Early Tamil-Brāhmī inscriptions is an orthographic feature which is discussed in Chapter 6.⁴

c25.s10.17

³ The occurrence of *ū* in the Early Tamil cave inscriptions is not noticed in A.H. Dani 1963 (Indian edition 1986): Pl. VIIa or in C. Sivaramamurti 1952: p. 65.

c25.s10.18

⁴ See section 6.15.

c25.s10.19

***e* / *ē*:** The letter *e* in Tamil-Brāhmī is rather unusual for its multiple geometric forms which defy any chronological arrangement. The variants with oblong, triangular, circular, semicircular and other odd shapes are illustrated in Chart 4: *e* / *ē*. The Early Vaṭṭeḻuttu form is arch-shaped (ibid. ***121*** ), very similar to the later forms.⁵

c25.s10.20

⁵ Cf. C. Sivaramamurti 1952: chart for *e* (Fig. 25, p. 68).

c25.s10.21

***o* / *ō*:** The letter *o*, formed by the addition of a short horizontal stroke at the top left of the vowel *u* occurs in angular (Chart 4: *o* / *ō*: ***18***) and cursive forms (ibid. ***6***). An exceptional variant (ibid. ***12***) is formed by the addition of the double strokes of the medial vowel *-o* at the top of the vertical of the initial vowel *u*.

c25.s10.22

**The short vowels *e* and *o*:** The length of the short vowels *e* and *o* is not indicated in the Early Tamil-Brāhmī inscriptions and has to be inferred from the context.⁶ The initial short vowel *e* with the *puḷḷi* placed inside the letter occurs in a Late Tamil-Brāhmī inscription (Chart 4: *e*:***85***) and also in the Early Vaṭṭeḻuttu inscriptions (ibid. ***117*** and Chart 8). It is also attested in a Tamil-Brāhmī coin-legend.⁷ No actual example of the initial short vowel *o* with the *puḷḷi* has been found so far.

c25.s10.23

⁶ This is what K.V. Subrahmanya Aiyer (1924: p. 283) meant when he stated that "in the case of combined consonants the occurrence of short *e* and short *o* deserves special attention, the two being the special characteristics of the Dravidian alphabets". He was not referring to the graphic forms as misunderstood by T. N. Subramanian (1938, reprint 1996: p. 22; 1957: p. 1507) and T.V. Mahalingam (1967: p. 134). For discussion on the contrast between *e*/*ē* and *o*/*ō* sounds in the Tamil-Brāhmī inscriptions, see section 6.16.

c25.s10.24

⁷ See No. 13, Table 1.6.

c25.s10.25

The unusual manner in which the letters of the short vowels *e* and *o* are formed from the respective long vowels deserves special attention. In the Brāhmī script, the letters of the long vowels are derived from those of the respective short vowels as in the pairs *a* > *ā*, *i* > *ī* and *u* > *ū*. The process is reversed in Tamil-Brāhmī only for the pairs *ē* > *e* and *ō* > *o*. In these cases, the Brāhmī script has only the long vowels. These were taken over by Tamil-Brāhmī and were employed, to begin with, to denote both the long and short sounds which could be recognised only from the context. However, as Tamil has also the short vowels *e* and *o*, the need to distinguish the length of the vowels was felt and the *puḷḷi* was added as a diacritic mark to the short vowels *e* and *o* (both initial and medial) to distinguish them from the respective long vowels. This development proves conclusively the derivation of Tamil-Brāhmī from the Brāhmī script.

c25.s10.26

***ai:*** The curious asymmetry in the occurrence of the vowel *ai* in Asokan Brāhmī¹ is also reflected in Tamil-Brāhmī. In both cases, the initial vowel is absent, but the medial vowel occurs from the earliest times.² The initial vowel *ai* is attested for the first time in the Early Vaṭṭeḻuttu inscription at Tirunatharkunru (Chart 4: *ai: 116*). The form of this letter looking like a 'trident' is quite different from the earliest form in Brāhmī.³ However, the trident-form of *ai* occurs in a Southern Brāhmī Prakrit inscription⁴ at an earlier date than in the Early Vaṭṭeḻuttu script.

c25.s10.27

¹ C.S. Upasak 1960: pp. 114-115.

c25.s10.28

² E.g., *tantai-y* 'father' (Corpus: No. 2). See also Palaeographic Chart 6: *-ai:* 2.

c25.s10.29

³ In the Hathigumpha inscription of Kharavela (K. G. Krishnan ed. 1989: No. 67). See illustration in A.H. Dani 1963 (Indian edition 1986): Table VI a. 9.

c25.s10.30

⁴ At Alluru (Krishna District., Andhra Pradesh) in *ARE* 331/1924. See illustration in T.N. Subramanian 1957: pp. 1518-1519; Pl. 2, Col. IV. 49.

c25.s10.31

***au:*** The vowel *au* (initial or medial) does not occur in Tamil-Brāhmī or Early Vaṭṭeḻuttu although it is included in the list of vowels in *Tolkāppiyam* (*Eḻu.* 4).

c25.s10.32

**The *āytam* (*ḵ*):** The unique Tamil sound *āytam* is not represented graphically in the Tamil-Brāhmī or Early Vaṭṭeḻuttu inscriptions.⁵ The phrase *muppāl puḷḷi* (*Tol. Eḻu.* 2) has been interpreted by some commentators (e.g., Nacciṉārkkiṉiyar) as 'dots on three sides' supposedly referring to the graphic shape of the *āytam*. However, as the character with three dots in a triangular pattern represents the initial vowel *i* in Tamil-Brāhmī, the same character could not have represented the *āytam*. The alternative interpretation of the *sūtra* as referring to 'three types of sounds marked by dots viz., the shortened *-i* and *-u* and the *āytam*' seems to be correct (e.g., Pērāciriyar). However, there is no inscriptional evidence that the dependent sounds, shortened *-i* and *-u*, were marked by the *puḷḷi*. It is likely that the present form of the *āytam* with three dots is itself a result of the later misinterpretation of this *sūtra* as referring to the graphic shape of the letter. There would have been no confusion as by this time in the early medieval period, the letter *i* was no longer being written as three dots (compare *i* in Charts 4 & 8).

c25.s10.33

⁵ The earliest inscriptional occurrence of the *āytam* is in the expression *veḵkā* 'name of a river' in the Kasakkudi Plates of Nandivarman II (ca. 753 A.D.). Here the letter is depicted as an upward-looking curve flanked by a pair of dots placed above and below (*SII*. II: No. 73, Pl. IXb, Line 115). This form is likely to have been influenced by the *visarga* symbol of the Grantha script (M. Raghawaiyangar 1938, reprint 1984: pp. 127-131). The *āytam* occurs in the expression *aḵkam* 'payment in grain' in the Vaṭṭeḻuttu inscription of Varaguna II (ca. 874 A.D.) at Tiruchendur. In this record, the letter is represented by a horizontal line with dots marked above and below 'like the division sign' (*EI.* XXI: p. 102).

5.11 Consonants

c25.s11.1

*k*: The simple form of *k*, a plain cross (Chart 2), remains stable with hardly any change throughout the period under study. There are, however, a few exceptional forms in Late Tamil-Brāhmī and Early Vaṭṭeḻuttu, which are influenced by the contemporary Southern Brāhmī forms (Chart 5A: *k*: *73, 83, 84, 85, 114 & 116*). These forms possess one or more of the following features of Southern Brāhmī, viz., head-marks (triangular or rectangular), drooping of the horizontal line and leftward curving of the lower end of the vertical. It is significant that these forms occur only in the northern and western districts adjoining Andhra and Karnataka. Another exceptional form is the conjunct consonant *kka* (ibid. *112*) occurring in the Early Vaṭṭeḻuttu inscriptions at Arasalapuram (Corpus: No. *112*, ca. 5th century A.D.) and Paraiyanpattu (ibid. No. *115*, ca. 6th century A.D.)

First 184 of 535 passages. Read on

Linked texts (14)