Pramāṇa · Search · About · API

Early Tamil Epigraphy: From the Earliest Times to the Sixth Century A.D. c25.s12: 5.12 Additional letters in Tamil-Brāhmī

Early Tamil Epigraphy: From the Earliest Times to the Sixth Century A.D. c25.s12

Language
English

Contents

5.12 Additional letters in Tamil-Brāhmī

c25.s12.1

The most interesting palaeographic feature of Tamil-Brāhmī is the formation of four additional letters to represent the consonants *ḷ*, *ḻ*, *ṟ* and *ṉ* in Tamil. It will be seen that in each case the additional letter is formed by a process of simple adaptation from the corresponding Brāhmī letter with the nearest phonetic value (Chart 3). The mode of formation of the letters provides valuable insight into their early phonetic affinities. These four letters serve as 'markers' of Tamil-Brāhmī and their presence in Southern Brāhmī and Sinhala-Brāhmī inscriptions of known dates provides new evidence for crosschecking the dates of the Tamil-Brāhmī inscriptions. The story of the discovery of the special letters in the Tamil cave inscriptions has already been narrated.¹ In this section, we shall study their origin and palaeographic evolution.

5.12.1 Tamil-Brāhmī ḻ

c25.s12.t1.1

(a) *Formation of* *ḻ

c25.s12.t1.2

The sound *ḻ* is produced when the tip of the tongue is raised and rubs the palate (*Tol. Eḻu.* 95). The sound -*ṭ*- (phonetically -*ḍ*-) alternates with *ḻ* in the Early Tamil-Brāhmī inscriptions.² The sound *ḻ* in Old Telugu becomes *ḍ* in later times.³ These linguistic correspondences find remarkable corroboration in the formation of Tamil-Brāhmī *ḻ* from Brāhmī *ḍ*, specifically from the Asokan form of *ḍ* (Chart 3:I.¹). Phonetically, *ḍ* appears to be quite close to the sound *ḻ*; (cf. Asokan *ēḍakē*⁴ with Ta. *ēḻakam* 'sheep').

c25.s12.t1.3

¹ See sections 2.3.1 & 2.4.2.

c25.s12.t1.4

² Corpus: Nos. 34 & 44. See Commentary (34.1.a). See also Bh. Krishnamurti 1958b on *ḻ* in Dravidian.

c25.s12.t1.5

³ K. Mahadeva Sastri 1969: p. 40.

c25.s12.t1.6

⁴ *CII.* I. Pillar Edict at Delhi-Mirath: V. 11.

c25.s12.t1.7

In Asokan Brāhmī, the letter *ḍ* is formed by the addition of a thick dot to the lower end of *ḍ*.⁵ The Tamil-Brāhmī *ḻ* is formed by transposing the dot (enlarged into a circle⁶) from the bottom to the top of the letter (Chart 3:I. 1 & 2). The rare Asokan *ḍ* is not known to have been in use after his time. This is important evidence that the origin of Tamil-Brāhmī cannot be dated much later.

c25.s12.t1.8

⁵ C.S. Upasak 1960: pp. 74-75; Appendix 19.

c25.s12.t1.9

⁶ G. Bühler 1896 (1959 reprint: p. 55) had suggested that the dot in *ḍ* might be a substitute for a circle.

c25.s12.t1.10

Dani derives Tamil-Brāhmī *ḻ* from Southern Brāhmī *ḷ*.⁷ His view is based on a comparison of the looped form of *ḻ* at Tirupparankunram with the form of *ḷ* occurring at Sanchi in ca.1st century A.D.⁸ Bühler derives *ḻ* from an even later prototype, the *ḷ* of the Gupta period.⁹ There is, however, hardly any resemblance between the two letters. In any case, we now know of the existence of much older forms of *ḻ* with the circular top (e.g., Corpus: Nos. *1-3*, ca. 2nd century B.C.) which bear no likeness to the Southern Brāhmī *ḷ*.

c25.s12.t1.11

⁷ A. H. Dani 1963 (Indian edition 1986): Fig. 7 (x) at p. 71.

c25.s12.t1.12

⁸ A.H. Dani (1986: p. 71) is mistaken in attributing the view that *ḻ* is derived from the Brāhmī 'double-looped' *ja* to K.V. Subrahmanya Aiyer. The comparison was made by H. Krishna Sastri (1919: p. 336, n. 7), which was not accepted by Subrahmanya Aiyer (1924: pp. 284-285) who emphasised the much closer likeness to the modern Tamil *ḻ*.

c25.s12.t1.13

⁹ G. Bühler 1896 (1959 reprint: p. 94; Pl. IV. 40. II).

c25.s12.t1.14

Figure 5.3. Pottery inscription from Kodumanal: use of *ḍ* as *ḻ* .

c25.s12.t1.15

Figure 5.4. Pottery inscription from Arikamedu: use of reversed *ḍ* as *ḻ*.

c25.s12.t1.16

Figure 5.5. Cave inscription in Sri Lanka with Tamil-Brāhmī *ḻ* (detail).

c25.s12.t1.17

Nagaswamy derives *ḻ* by combining the Brāhmī letters *ṭh* and *ḍ*.¹ However, the suggested derivation is doubtful as it is unlikely that a simple consonant would be derived by combining two other consonants.

c25.s12.t1.18

¹ R. Nagaswamy 1979: p. 75.

c25.s12.t1.19

Karunaratne derives the Tamil-Brāhmī *ḻ* from the reversed form of Brāhmī *ḍ* by adding hypothetically a small loop in the middle.² However, the evolution of the looped *ḻ* is better explained morphologically as a later development from the older circular form when written with a running hand (cf. Chart 3: I. 3 & 4; Chart 5B: *ḻ*:55 & 76).

c25.s12.t1.20

² S. Karunaratne 1984: pp. 33-35.

c25.s12.t1.21

(b) *Evolution of ḻ

c25.s12.t1.22

In Early Tamil-Brāhmī, *ḻ* has a circular top which may or may not be bisected by the vertical (Chart 3: I . 2 & 3; Chart 5B:*ḻ*:17 & 49). Both variants are also attested in the pottery inscriptions from Kodumanal (Chart 7: KDM 27 & 164; 148 & 3). The step-like line derived from Brāhmī *ḍ* (Corpus: *ḷu* in No. *1* and *ḷi* in No. 2), tends to change from the angular to cursive shape (Chart 5B: *ḻ*: 17). Towards the end of the Early Period, the top circle opens out and forms a large loop with a gracefully curved tail as at Tirupparankunram (Chart 3 : I.⁴; Chart 5B : *ḻ*: 55). Still later, the tail gets straightened (Chart 3: I .⁵; Chart 5B : *ḻ*: 77). It is this late form that is seen in the pottery inscriptions at Arikamedu (Chart 7: AKW:16). The Early Vaṭṭeluttu form is similar but somewhat more cursive (Chart 3 : I.⁶; Chart 5B : *ḻ*: 116).

c25.s12.t1.23

(c) *Use of ḍ as ḻ in pottery inscriptions from Kodumanal

c25.s12.t1.24

A unique feature of the pottery inscriptions from Kodumanal³ is the apparent use of the Brāhmī letter *ḍ* to represent the sound *ḻ*. Apart from the forms of Tamil-Brāhmī *ḻ* with the circular top,⁴ there are five occurrences of *ḍ*,⁵ a letter which is not attested elsewhere in Tamil-Brāhmī inscriptions. Even though the pottery inscriptions are too fragmentary, it seems unlikely that inscriptions with the letter *ḍ* are in Prakrit, as almost all the inscriptions at this site are in Tamil. It is also unlikely that -*ḍ*- represents the voicing of -*ṭ*- in medial positions as such a usage is unknown in Tamil-Brāhmī inscriptions at this site⁶ or elsewhere.⁷ The conclusion, therefore, appears inescapable that there was a unique innovation in the pottery inscriptions at this site, which permitted the employment of the Brāhmī *ḍ* to represent the Tamil sound *ḻ*. A plausible example is the word ...*toḷut*[*t*]*ai* 'slave woman, servant maid' (*TL*), where *ḍu* seems to have the value *ḻu* (Fig. 5.³).¹ This tentative suggestion is supported by the evidence regarding the use of the reversed *ḍ* to represent *ḻ* at Arikamedu and in the loanwords in the cave inscriptions of Sri Lanka, which we shall consider next.

c25.s12.t1.25

³ Y. Subbarayalu, Catalogue of Pottery Inscriptions from Kodumanal 1996 (unpublished).

c25.s12.t1.26

⁴ The circular forms of *ḻ* with bisected top circle occur in KDM 3 & 148, and without bisecting in KDM 27, 84 & 164 (see Palaeographic Chart 7 for examples).

c25.s12.t1.27

⁵ The Brāhmī letter *ḍ* probably stands for the sound *ḻ* in KDM 2, 70, 83?, 151 & 167. (See Palaeographic Chart 7: KDM 70 & 2 for examples.)

c25.s12.t1.28

⁶ E.g., asaṭa[*ṇ*] and neṭuñ-, where the medial -*ṭ*- occurs (KDM 38 & 110 respectively).

c25.s12.t1.29

⁷ See section 6.22 for discussion on voicing of consonants in Tamil-Brāhmī.

c25.s12.t1.30

¹ KDM 2. Y. Subbarayalu (1988:p. 120) reads *...toṭittai*.

c25.s12.t1.31

(d) *Use of reversed ḍ for ḻ in pottery inscriptions from Arikamedu

c25.s12.t1.32

Another connected innovation is the representation of the sound *ḻ* by the reversed form of *ḍ* in the pottery inscriptions at Arikamedu. The choice of the reversed form of *ḍ* (unlike at Kodumanal) has the advantage of avoiding confusion with Prakrit inscriptions which also occur minimally along with Tamil-Brāhmī inscriptions on pottery at ancient Tamil sites. Karunaratne² first pointed out that in a Prakrit inscription from Arikamedu,³ the reversed *ḍ* denotes the Tamil *ḻ* in the expression *yaḻu-valabhutaya*⁴ 'of Yaḻu Balabhūtā'.

c25.s12.t1.33

² S. Karunaratne 1984: p. 34.

c25.s12.t1.34

³ K.A. Nilakanta Sastri 1942: pp. 3-4 and Fig. 5. P. Z. Pattabiramin 1946: Pl. XXI, No. 3. R.E.M. Wheeler *et al.* 1946: p. 114, Fig. 47. 20. I. Mahadevan 1973: pp. 60-64, Inscription No. IV, Fig. 4.

c25.s12.t1.35

I have found another instance in a Tamil-Brāhmī pottery inscription (Fig. 5.⁴) at Arikamedu from the excavations by Wheeler and published by him but not read satisfactorily so far.⁵ The earlier part of the inscription is fragmentary with portions missing; the later part is clear; it reads:

c25.s12.t1.36

⁴ In my 1973 paper, I had read the expression as *yaḍu-valabhuti-y* on the basis of the eye copy published by Pattabiramin and interpreted the word as a 'masculine' personal name and hence regarded the language as Tamil. I have reconsidered the matter after re-inspection of the original sherd and also in the light of the new evidence presented here. I now consider this to be a Prakrit inscription influenced by Tamil. See the previous two notes for references.

c25.s12.t1.37

⁵ R.E.M. Wheeler *et al.* 1946: p. 112, Fig. 46, No. 4 (right fragment).

c25.s12.t1.38

. . . [*ma*] *ṇa ca ṇ ṭai kō va ḻā ṉ* 'Aḻāṉ, the chief of . . . maṇacaṇṭai '.

c25.s12.t1.39

The penultimate letter, perfectly preserved, is the reversed form of *ḍā* with the value *ḻā* which fits the context.⁶

c25.s12.t1.40

⁶ *kōvaḻāṉ* is analysed as *kō* 'chief' + *v* (glide) + *aḻāṉ* (name). The name is attested in Old Tamil; cf. *aḻāṉ* and *aḻāntai*, cited as illustrations in *Tol. Eḻu.* 349, comm., Iḷam. The name *aḻāṉ* is also attested as a variant reading in *Naṟṟ.* 121, comm. Mayilai. The incomplete expression *... maṇacaṇṭai* is probably a place name.

c25.s12.t1.41

The open looped form of *ḻ* with the straight tail (a late feature) also occurs at Arikamedu (Chart 7: AKW 16). The earlier circular form of *ḻ* is not attested at this site.

c25.s12.t1.42

(e) *Use of reversed ḍ for ḻ in Sinhala-Brāhmī inscriptions

c25.s12.t1.43

Karunaratne has drawn attention to the employment of the reversed *ḍ* of Brāhmī with the sound *ḻ* in Early Sinhala-Prakrit inscriptions.⁷ He points out that in the expression *damēḍa*⁸ (corresponding to Ta. *tamiḻa*), the letter *ḍ* is reversed even though the direction of writing of the inscription is not reversed. His inference that the reversed form of *ḍ* in such cases has to be read as *ḻ* is acceptable.

c25.s12.t1.44

⁷ S. Karunaratne 1984: pp. 33-35; Fig. 35

c25.s12.t1.45

⁸ Ibid. Inscription No. 48.

c25.s12.t1.46

However, his suggestion that the reversed *ḍ* found in Sri Lanka is "the earliest known form of Dravidian *ḻa*"¹ needs to be reconsidered in the light of the new evidence on the formation and evolution of Tamil-Brāhmī *ḻ*.

c25.s12.t1.47

¹ S. Karunaratne 1984: pp. 34-35.

c25.s12.t1.48

(f) *Occurrence of Tamil-Brāhmī ḻ in a Sinhala-Brāhmī inscription*

c25.s12.t1.49

I have also found a unique and hitherto unnoticed occurrence of the early circular form of the Tamil-Brāhmī letter *ḻ* (but with reversed orientation) from the published estampages² of an early cave inscription at Bambaragastalava, Eastern Province, Sri Lanka (detail in Fig. 5.5). The expression in which this form occurs is a personal name which has been read as *aśaḍiya* by Paranavitana and as *aśaḷiya* by Karunaratne, both treating the third letter as the reversed form of *ḍi*.³ However, these expressions occur nowhere else.⁴ The letter in question which is clearly seen in the estampages in both the publications, is the earlier circular form of Tamil-Brāhmī *ḻa* but with reversed orientation and the top circle (looking squarish as in Kodumanal)⁵ bisected by the vertical line. I read the expression as *aśaḻaya* (*aśaḻa* + *aya* < Pkt. *asaḍa* + *aya* < Skt. *āshāḍha* + *ārya*).⁶ This unique occurrence confirms the existence of the Tamil-Brāhmī script from at least the 2nd century B.C., the date assigned to this inscription by Karunaratne.

c25.s12.t1.50

² S. Paranavitana 1970: No. 504. S. Karunaratne 1984: No. 15.

c25.s12.t1.51

³ S. Paranavitana 1970: No. 504. S. Karunaratne 1984: No. 15. See also Karunaratne's discussion on this letter at p. 33. His eye copies of the letter at p. 33 and in Fig. 35 are incorrect. The reading *aśaḷiya* at p. 58 appears to be a typographic error.

c25.s12.t1.52

⁴ Paranavitana (1970: Glossary) lists *asaliya*, the genitive singular of *asali* (< Skt. *āślēsha*). But the letter under discussion here cannot be read as *li*.

c25.s12.t1.53

⁵ Cf. *Palaeographic Chart 7*: *ḻi*: KDM 164.

c25.s12.t1.54

⁶ Cf. *aśaḍa* and the suffix *-(a)ya* in Sinhala-Prakrit inscriptions (S. Paranavitana 1970: Glossary). Interestingly, the Tamil form of the name occurs as *asaṭa[n]* in a pottery inscription from Kodumanal (KDM 38).

c25.s12.t1.55

(g) *Occurrence of Tamil-Brāhmī ḻ in a Southern Brāhmī inscription*

c25.s12.t1.56

In 1942, Nilakanta Sastri⁷ published a Prakrit inscription of the Ikshvāku dynasty found earlier at Gurzala near Nagarjunakonda. The record is in the Southern Brāhmī script of the 3rd century A.D. and gives the name of the ruler as *siri-ruḷu-purisa-dāta*-as read by Nilakanta Sastri. (Later, Krishnan⁸ corrected the reading *-ruḷu-* as *-ruḻa-* <Pkt. *ruḍa* < Skt. *rudra*.) Interestingly, when Nilakanta Sastri published this inscription, the occurrence of a similar circular form of *ḻ* in the Tamil cave inscriptions had not been recognised; he identified the letter *ḻ* by comparing it with a somewhat similar form found in a later Western Chalukya record of the 7th century A.D.⁹

c25.s12.t1.57

⁷ *ARE* 335/1936-37. K.A. Nilakanta Sastri 1941-42: pp. 123-125 (with Pl.).

c25.s12.t1.58

⁸ K.G. Krishnan (ed.) 1989: p. 457, n. 5.

c25.s12.t1.59

⁹ Badami Rock Inscription of Mangalesa (597-610 A.D.) in *IA*. X, Pl. facing p. 59, Line 4.

c25.s12.t1.60

Figure 5.6. Forms of *ḻ* in Early Southern scripts.

c25.s12.t1.61

Figure 5.7. Cave inscription in Sri Lanka with Tamil-Brāhmī *ṟ* (detail).

c25.s12.t1.62

Figure 5.8. Cave inscription in Sri Lanka with Tamil-Brāhmī *ṉ* (detail).

5.12.2 Tamil-Brāhmī ḷ

c25.s12.t2.1

(a) *Formation of ḷ

c25.s12.t2.2

The letter *ḷ* of Tamil-Brāhmī is formed by attaching a small angular stroke (looking like an inverted L) to the letter *l* (Chart 3:II. 1 & 2). The same attachment is employed for deriving *ṟ* from *ṭ* and *ṉ* from *n* indicating that these letters were formed at the same time when Tamil-Brāhmī was adapted from the Brāhmī script. The angular stroke is also drawn cursively from the earliest times (ibid. II.³).

c25.s12.t2.3

(b) *Evolution of ḷ

c25.s12.t2.4

In course of time, the attached stroke at the right of the letter moves downwards (Chart 3:II.⁴) or upwards (ibid. II. 5). In Late Tamil-Brāhmī, the right side of the letter and the attachment merge into a single vertical line (ibid. II.⁶). In Early Vaṭṭeḻuttu, there is a change of direction (as in *l*) and the letter is written from the left as a continuous curving line (ibid. II.7). See also Charts 5B and 6 to 8 for attested forms.

c25.s12.t2.5

(c) *Occurrence of Tamil-Brāhmī ḷ in Early Sinhala-Brāhmī inscriptions

c25.s12.t2.6

Parker's identification of *ḷ* in the Early Sinhala-Brāhmī inscriptions was disputed by Paranavitana¹ even after the correctness of the identification was proved beyond doubt by the occurrence of the letter in appropriate linguistic contexts in the numerous Tamil cave inscriptions discovered since then. Paranavitana's laboured attempt to read the letter as *lu* is unconvincing, as forms like *ḷi* and *ḷe* occurring in the Sri Lankan cave inscriptions clearly indicate that the basic form must be a simple consonant.² The loanwords from Tamil occurring in these inscriptions can be read meaningfully only if the letter in question is treated as *ḷ*.³ Karunaratne accepts the value *ḷ* and records its occurrence in five early inscriptions assigned by him to the 2nd century b.c.⁴ He has also pointed out that this form of *ḷ* was later replaced by the form in vogue in the Brāhmī inscriptions in India. The occurrence of Tamil-Brāhmī *ḷ* in the Early Sinhala-Brāhmī script proves the contemporaneity of the two scripts.

c25.s12.t2.7

¹ S. Paranavitana 1970: pp. xxiii-xxv.

c25.s12.t2.8

² Ibid. *ḷi* (No. 78), *ḷe* (No. 967).

c25.s12.t2.9

³ Ibid. e.g., *vēḷ* (No. 123), *parumakaḷ* (No. 331), *iḷa-* (No. 94) from Ta. *iḷa-* or *īḻa-*.

c25.s12.t2.10

⁴ S. Karunaratne 1984: pp. 32-33; Fig. 34; Inscriptions 8, 18, 27, 29 & 31. S. Paranavitana (1970: p. xxv) gives a list of 23 inscriptions where this letter occurs. A.H. Dani (1963, Indian edition 1986) does not include this form in his Table (ibid. p. XIX b) which lists only the Southern Brāhmī form of *ḷ* occurring in Late Sinhala-Brāhmī.

c25.s12.t2.11

(d) *Occurrence of Tamil-Brāhmī ḷ in a Southern Brāhmī inscription

c25.s12.t2.12

The hill at Malakonda, Nellore (now Prakasam) District in Andhra Pradesh, is strategically located to the north of the traditional boundary between Andhra and the ancient Tamil country, and probably also marked the southern limit of the Mauryan Empire in this region. The cave inscription⁵ on this hill is in Prakrit written in the Southern Brāhmī script of ca. 1st century a.d.⁶ and is the southernmost example of its kind.¹ The inscription records the gift of the cave shelter presumably to Jaina monks by a Tamil trader (*seṭhi*) who describes himself as belonging to the *aruvāḷa-kula*.² The Tamil-Brāhmī form of *ḷ* occurs uniquely in this expression.

c25.s12.t2.13

⁵ *ARE* 531/1937-38 (with illustration). Also in T.V. Mahalingam 1967: p. 308, Pl. 30.

c25.s12.t2.14

⁶ The inscription is dated in the 3rd century b.c. in *ARE* (ibid.). However, the developed forms of letters like the square *pu* and the triangular *va* indicate a date around the 1st century a.d.

c25.s12.t2.15

(e) *Different forms of ḷ in Early Southern scripts

c25.s12.t2.16

It is interesting that there are three distinct forms of *ḷ*, one each in the Tamil-Brāhmī, Southern Brāhmī and Bhattiprolu scripts (Fig. 5.⁶, a to c respectively). Though the sounds are identical or nearly so, the three forms of *ḷ* appear to be derived from different sources.³ As we have seen above, the Tamil-Brāhmī *ḷ* is derived from *l*. The Southern Brāhmī *ḷ* was developed from the later cursive form of Brāhmī *ḍ* (with a right-facing curve) by adding a small semicircle at the lower end.⁴ The form of *ḷ* at Bhattiprolu cannot be related to any other letter and its origin is unknown; it was probably a short-lived local experimentation.⁵

c25.s12.t2.17

³ C. Sivaramamurti (1952: chart for *ḷ* in Fig. 64 at p. 152) connects the three forms of *ḷ* in a single evolutionary sequence; but the oldest forms in his own chart do not support the arrangement.

c25.s12.t2.18

⁴ G. Bühler 1896 (1959 reprint): p. 30. Dani's hypothetical derivation of the Southern Brāhmī *ḷ* from *ṭ* is improbable (A.H. Dani 1963, Indian edition 1986: p. 54, Fig. 6 viii). Phonetically and graphically, the sound *ḷ* is closer to *ḍ* as pointed out by Bühler.

c25.s12.t2.19

⁵ Bühler 1894b (*EI.* II:p. 324) derives the Bhattiprolu *ḷ* from the Sanchi *ḷ*. Dani (1986: p. 71, Fig 7 vi) derives the Bhattiprolu *ḷ* from the Brāhmī *l*.

5.12.3 Tamil-Brāhmī ṟ

c25.s12.t3.1

(a) *Formation of ṟ

c25.s12.t3.2

The Dravidian sound *ṟ* (<PDr *\* ṯ*) was originally an alveolar stop which later developed into a trill in Tamil (except in the Jaffna dialect).⁶ It has become almost indistinguishable from *r* in modern spoken Tamil. However, it is described as a *valliṉam* (stop) in Tamil grammatical tradition (*Tol. Eḻu.* 19) and, like other stops (and unlike *r*), cannot occur word-finally.

c25.s12.t3.3

⁶ Kamil Zvelebil 1970: p. 96.

c25.s12.t3.4

The Tamil-Brāhmī letter *ṟ* is formed from *ṭ* (Chart 3:III.¹) to which a short vertical line is added at the bottom (ibid. III.²) to support the angular attachment added to the left or right of the vertical (ibid. III. 3 & 4). The choice of *ṭ* to derive the Tamil-Brāhmī *ṟ* is significant as it indicates that the sound *ṟ* was still an alveolar stop (nearer to *ṭ* than to *r*) in the language of the cave inscriptions.⁷

c25.s12.t3.5

¹ All early cave inscriptions in the Peninsula to the south of this site are in Tamil written in Tamil-Brāhmī.

c25.s12.t3.6

² The Aruvāḷar are mentioned in Caṅkam literature as the Tamil tribe which inhabited the northern border regions of ancient Tamil country, cf. *Paṭṭiṉa.* 275; *Tol. Pāyiram*, comm. Nacc.

c25.s12.t3.7

⁷ It is significant that there is no instance of *r* occurring in the place of *ṟ* or *vice versa* in the Tamil-Brāhmī inscriptions included in the Corpus.

c25.s12.t3.8

Dani's hypothetical derivation⁸ of *ṟ* from *ḷ* at Sanchi ignores the phonetic distinction between the two sounds and is based on a relatively later form of *ṟ*. He is unable to recognise the earlier forms of *ṟ* which he misreads as *ṭu* or *ṭū*.⁹

c25.s12.t3.9

⁸ A. H. Dani 1963 (Indian edition 1986): pp. 70-71, Fig. 7 (ix).

c25.s12.t3.10

⁹ Ibid. Pl. VII a, Nos. 7 & 9.

c25.s12.t3.11

According to Meenakshisundaran¹ and Nagaswamy,² the letter *ṟ* is formed by combining the letters *ṭ* and *t*. However, the form of *ṟ* illustrated by Nagaswamy is a later one. Further, Tamil-Brāhmī has no ligatures and it is very doubtful whether in this script a simple consonant would be derived from a ligature of two other consonants. A comparison with the mode of formation of the other three additional letters (*ḻ*, *ḷ* and *ṉ*) clearly indicates a similar derivation for *ṟ* also by the addition of a simple diacritic-like attachment.

c25.s12.t3.12

¹ T. P. Meenakshimsundaran 1965: p. 41.

c25.s12.t3.13

² R. Nagaswamy 1979: pp. 74-75.

c25.s12.t3.14

(b) Evolution of ṟ

c25.s12.t3.15

In the earliest forms of *ṟ* (Chart 3: III. 3 & 4), the constituent parts viz., the basic letter *ṭ*, the lower vertical and the angular attachment are all distinctly seen. Later, however, the vertical slants to the left and the angular attachment is abbreviated to a single stroke slanting to the right with the result that the two elements together look like the letter *t* joined to the letter *ṭ* (ibid. III.⁵).³ In another parallel development seen in the pottery inscriptions at Kodumanal, the lower vertical alone survives without the angular attachment or additional stroke (Chart 7: KDM 120 & 89). It is this form which leads to the more cursive shape seen in Early Vaṭṭeḻuttu (Chart 3: III.⁶). See also charts 5B and 6 to 8 for attested forms.

c25.s12.t3.16

³ It is this late form which led to the suggestion that the letter *ṟ* is a combination of *ṭ* and *t* as discussed above.

c25.s12.t3.17

(c) Occurrence of Tamil-Brāhmī ṟ in a Sinhala-Brāhmī inscription

c25.s12.t3.18

Pushparatnam,⁴ has recently located the occurrence of the Tamil-Brāhmī letter *ṟ* in one of the Sinhala-Brāhmī inscriptions published by Paranavitana (detail in Fig. 5.7).⁵ The letter occurs in the expression *maṟa(ru)makaṉ* (read by Paranavitana as *marumakana-*).⁶

c25.s12.t3.19

⁴ P. Pushparatnam 2000b (unpublished): A survey of the occurrence of the kinship term *marumakaṉ* in the Brāhmī Inscriptions of Sri Lanka, Fig. 4.

c25.s12.t3.20

⁵ S. Paranavitana 1970: No. 487 (1) and estampage in Pl. LII. Note especially the angular attachment at lower right clearly identifying the letter as Tamil-Brāhmī *ṟa*.

c25.s12.t3.21

⁶ Probably from Ta. *marumakaṉ* lit., 'nephew', also used as a term of endearment.

5.12.4 Tamil-Brāhmī ṉ

c25.s12.t4.1

(a) Formation of ṉ

c25.s12.t4.2

The Tamil-Brāhmī *ṉ* is formed by the addition of an angular or cursive attachment to the top of the Brāhmī letter *n* (Chart 3: IV.1-3).

c25.s12.t4.3

Dravidian linguists are not agreed on the question whether the dental nasal *n* and the alveolar nasal *ṉ* are two different phonemes or merely allophones of one phoneme. These are, however, treated as different sounds in Tamil grammatical tradition (*Tol. Eḻu.* 20). The fact that those responsible for the invention of the Tamil-Brāhmī script fashioned a separate character for *ṉ* by modifying *n* is clear evidence that there must have been a distinction in sound between the two letters in the language at that time. The retention of two sounds (but represented by one letter) in Malayalam even at present corroborates the view that the alveolar nasal *ṉ* was a distinct phoneme in Old Tamil. However, even in the cave inscriptions there are instances of *ṉ* being written as *n* (e.g., No. 59).

c25.s12.t4.4

(b) *Evolution of ṉ

c25.s12.t4.5

The earliest form of *ṉ* in the cave inscriptions resembles an inverted J (Chart 3: IV.³). The angular form (ibid. IV.²) preserved in the pottery inscriptions at Arikamedu (Chart 7: AKW 9; AKB 22) and Kodumanal (ibid. KDM 38) is probably equally early as indicated by the similar mode of formation of the letters *ḷ*, *ṟ* and *ṉ*. In one of the variant forms, the top attachment is almost flattened making the letter look like *nā* (Chart 5B: *ṉ*: *21*). In course of time, the baseline to the left of the vertical becomes shorter and the top hook gets bigger and tends to bend lower until it almost touches the baseline and goes down even below (Chart 3: IV. 4). In a parallel development, the baseline gets elongated towards the left and the hook at right becomes smaller (ibid. IV. 5). In Early Vaṭṭeluttu, the baseline has shrunk into a small hook attached to the top left of the larger curve (ibid. IV. 6). See also Charts 5B and 6 to 8 for attested forms.

c25.s12.t4.6

(c) *Occurrence of Tamil-Brāhmī ṉ in Sinhala-Brāhmī inscriptions

c25.s12.t4.7

Pushparatnam,¹ has recently discovered the occurrence of the Tamil-Brāhmī letter *ṉ* in a couple of Sinhala-Brāhmī inscriptions published by Paranavitana (e.g., detail in Fig. 5.8).² He made the discovery in the course of his study of the Tamil loanword *marumakaṉ* occurring in the cave inscriptions of Sri Lanka. While in most cases the loanword is spelt with the dental *n*, the alveolar nasal *ṉ* of Tamil-Brāhmī occurs twice.³ This discovery is important because the expression is shown to be borrowed from Tamil,⁴ and also because the occurrence of *ṉ*, the most characteristic letter of Tamil-Brāhmī, in the Sinhala-Brāhmī inscriptions proves the existence of the Tamil-Brāhmī script from at least the 2nd century B.C., the date to which one of these inscriptions (No. 643) may be assigned on palaeographic evidence.

c25.s12.t4.8

¹ P. Pushparatnam 2000b (unpublished): Figs. 1 & 2.

c25.s12.t4.9

² S. Paranavitana 1970: No. 643 (Early Period) and No. 1161 (Late Period).

c25.s12.t4.10

³ Ibid.

c25.s12.t4.11

⁴ Cf. especially the variant *marumān* < Ta. *marumakaṉ* in No. 1161 (ibid.).

Linked texts (7)