Home | Biodata | Biography | Photo Gallery | Publications | Tributes
[Back to Epigraphy List]


On the occurrence of the pulli in the Tamil-Brahmi inscription of A$naimalai
 New Dimensions in the Study of Tamil Culture, 60th Birthday Felicitation Volume of Prof.V.Vanamamalai, Palayamkottai, 1978, pp. 8-12
STAT-34/78, January 1978
A paper presented at the Fourth Annual Congress of the Epigraphical Society of India at Madras, January 1978.
Emmanuel Jebarajan and Gift Siromoney 


The majority of early Tamil inscriptions written in a script similar to the Asokan Brahmi script have been found on small hills in the Madurai region. Two notational systems of  writing have been distinguished and both these systems differ from the Tamil Pulli system described in Tolka$ppiam.

Rock inscriptions using the Tamil Pulli system have been noticed at Pugalu$r and Araccalu$r which are situated at about 150 kms north of Madurai. It has been held that these inscriptions are later than the Tamil-Brahmi inscriptions of the Madurai region.

The existence of a Tamil-Brahmi inscription at A$naimalai, a suburb of Madurai, has been known for many decades. We show that the Tamil Pulli system is followed in this well-preserved rock inscription. A pulli is found as an unmistakable circular depression next to the letter t@. The occurrence of the pulli in the A$naimalai inscription shows that the Tamil Pulli notational system existed at a date earlier than the one hitherto accepted by epigraphists.

Some questions on the methods of dating Tamil-Brahmi inscriptions are also raised.

Early Tamil inscriptions written in characters closely resembling the letters of the Asokan Brahmi script came to light at the beginning of the century. A great impetus was given to their study in the 1960's by I.Mahadevan who successfully deciphered a large number of these inscriptions. Tamil characters have been found in rock shelters, on potsherds of different types1 (including black-and-red ware2 and rouletted ware3), on a silver coin of a Satavahana king4 of the second century A.D. and on the lead coins of Sendan, found in the Chengam hoard.

Mahadevan distinguished two notational systems in these writings. In the first system, which we shall call the Tamil-Brahmi system, the short a (medial) was denoted by attaching a short horizontal stroke to the middle or the upper right side of a consonant symbol. The lack of such a vowel-marker denoted the pure consonant. Mahadevan also found a second system, which we shall call the Brahmi system, in which the pure consonant as well as the consonant with the inherent a (short medial) were represented by the same symbol as in Asokan Brahmi. The horizontal stroke represented the long a (medial). In addition to these two systems there is also the Tamil Pulli system described in Tolka$ppiam where a. pure consonant is distinguished by a pulli, a consonant-marker. The pulli was also used in that system to shorten the vowels e and o.

At the International Tamil Conference held in 1968 at Madras, Mahadevan declared that the pulli did not occur in any of the cave inscriptions.5  In a series of lectures delivered at Madurai in 1970 he made out that the Tamil Pulli system is a later development of the Brahmi system and claimed that it had developed around 200 A.D.6 By then it had been well-established that the bilingual silver coin of a Satavahana king of the second century A.D. had a pulli to denote a pure consonant. The lead coins of Sendan also had a pulli to denote the short e.

If one were to accept the view that the pulli was invented around 200 A.D. and not earlier, then Tolka$ppiam itself will have to be assigned to a period around 200 A.D. This view is in conflict with the generally held view that Tolka$ppiam is the oldest of the Sangam works. Kurunthogai which is admittedly later than Tolka$ppiam is assigned a date of 180 B.C.7 The dates for Tolka$ppiam arrived at by an epigraphist and a linguist differ by as much as 400 years. This is far from satisfactory.

A$naimalai (elephant hill) which resembles a huge reclining elephant, is located in a suburb of Madurai and the existence of a Tamil-Brahmi inscription there has been known for many decades. The inscription is one of the best preserved of the cave inscriptions8 but scholars have failed to notice any evidence of the occurrence of the pulli there. In the inscriptions of Pugalu$r and Araccalu$r situated at about 150 kms north of Madurai, the occurrence of the pulli has been noticed in the recent past and A$naimalai inscription has been considered to be earlier than these two inscriptions.

On close examination of the A$naimalai inscription, one can clearly see the unmistakable existence of a pulli in the word a ra  t@  t@a (Figure 1). It is inscribed in the form of a circular dot next to the letter t@ right in the middle of the line and not above the line. This is similar to the position of the pulli in the letter t@ in the Satavahana coin. At Anaimalai the pulli is a little outside the semicircular shape of the letter t@. This is perhaps the reason why it had not been noticed by others. Its circular shape may be contrasted with the shape of the pulli in the Satavahana coin where it is represented as a short horizontal stroke. In both the coin and the cave inscription the consonant t@ is doubled.

The word preceding the word a  ra  t@@@  t@a is a  t  tu  va$  yi. A pulli is observed next to the letter t@ as can be seen from the estampage (Figure 1) but this cannot be asserted as strongly as the previous case.

The inscription is read using the second notational system and the discovery of the pulli does not alter the reading. This is the first time that the Tamil Pulli system has been found to be in use in the Madurai region. Mahadevan has assigned the A$naimalai inscription to the period 1-2 century A.D. since it follows the notational system of Arikamedu graffiti. The discovery of the occurrence of the pulli makes it necessary to either assign the inscription to a latter date or assign an earlier date to the use of the Tamil Pulli notational system than Mahadevan's suggested date of 200 A.D.

The methods that have been used so far to date the Tami -Brahmi inscriptions leave some questions unanswered.

If the first system which we call the Tamil-Brahmi notational system was specially adapted from Asokan Brahmi for the Tamil language then why does the second system which is closer to the Asokan Brahmi system replace the first system? Is it not because the second system is as good if not better than the first system for writing Tamil?

If the first system is decidedly older than the second system then why does not Tolka$ppiam mention it? What is the basis for saying that the linguistic features of those inscriptions that use the first system are older than the features of inscriptions that follow the Tamil Pulli system? Is not Tolka$ppiam used as a basis for making such assertions? 

The writings on Uraiyu$r potsherds are found to be similar, with reference to the shape of the letters, to the graffiti found at Arikamedu, and the Uraiyu$r graffiti are given the same date as the Arikamedu graffiti. If this is acceptable then how does one explain the fact that the Uraiyu$r writings follow the first notational system and the Arikamedu second? Did the two systems exist side by side?

The Arikamedu graffiti are assigned to the period of 1-2 century A.D. by Wheeler on the basis of the occurrence of Italian Arretine ware. But he also admits that Arretine ware had been manufactured with the same markings from the beginning of the first century B.C. to the middle of the first century A.D. Wheeler himself admits that "the evidence is consistent with an earlier initial date for the Arikamedu series".9 Therefore how much reliance should one place on Wheeler's initial date for the Arikamedu graffiti?

How reliable is the method based on paleography10 for determining the dates of Tamil-Brahmi inscriptions? What are the period characteristics and what are the scribe characteristics? Are there differences between the writings of rural and urban scribes? If one takes the extension of the vertical line of the letter ka below the lower line of the letter n^a, as well as the thickening of the top portion of the vertical line as period characteristics then A$naimalai inscriptions must belong to a period anterior to the Satavahana bilingual coin of the end of the second century A.D. These characteristics are found in the Araccalu$r inscriptions. However Araccalu$r is dated to a later period11 by an eminent scholar. Another eminent scholar first assigned the same inscription to the end of the second century A.D.12 and a few years later to the 3-4 century A.D.13

Do the inscriptions, in general follow consistently any one system? The letter ka is often represented without the horizontal vowel-marker in many inscriptions which are supposed to use the first system. Even in the Satavahana coin there is lack of consistency. In this coin the first notational system is not followed. However the letter ma is represented with a horizontal vowel-marker which would make the letter ma$. We read it as ma as though the system is the first system and we use it only for this letter in the legend. How much reliance can be placed on the assumption that the scribe followed a single system?

 How do the scholars who use the method of paleography arrive at dates which differ by as much as two centuries? Ma$mandur inscription is assigned to the first century A.D.14 by Nagaswamy and to the 3-4 century A.D.15 by Mahadevan? Should any good system of dating allow for an error of two or three centuries?

In Mahadevan's scheme of three systems of Tamil-Brahmi, Brahmi and the Tamil Pulli system, pulli does not occur in the first two systems. By careful study we have shown the occurrence of the pulli in one inscription that was classified as belonging to the second system. Are there any more inscriptionssuch as Alagarmalai inscriptionsof the second system which also have the pulli? Does the pulli also occur in any of the inscriptions which have been classified as belonging to the first system?

These questions show a great need for further study in this area.

The authors wish to thank Dr Michael Lockwood and Dr. K.V. Raman for fruitful discussions.

Figure 1 . Fragment of A$naimalai inscriptions showing the occurrence of the pulli in the second line. The white spot in the first line is not a pulli.


1. S. Gurumurthy (1976), Inscribed potsherds from South Indian excavations, Studies in Indian Epigraphy, Vol. Ill 120-123.
2. T.V. Mahalingam (1968), Inscribed potsherds from Alagarai and Uraiyu$r, Seminar on Inscriptions, Madras, 42-43.
3. A. Swamy (1973), A rare potsherd (rouletted ware) inscription from Arikamedu, Bulletin of the Institute of Traditional Cultures, Madras, 78-90.
4. R. Nagaswamy (1966), A bilingual coin of Satavahana, Seminar on Inscriptions, 200-202.
5. I. Mahadevan (1968), Tamil-Brahmi inscriptions of the Sangam age, Plenary Sessions, II International Conference-Seminar of Tamil Studies, Madras.
6. I. Mahadevan (1970), Tamil Brahmi Inscriptions, Madras.
7. V.I. Subramoniam (1970), The dating of Sangam literature, Proceedings of the Third International Conference Seminar, Paris, 75-90.
8. T.V. Mahalingam (1974), Early South Indian Paleography, Madras.
9. R.E.M. Wheeler (1946), Arikamedu, Ancient India, No. 2, p. 40.
10. K.V. Ramesh (1976), Paleography in dating undated inscriptions, Studies in Indian Epigraphy, Vol. Ill, 156-162.
11. I. Mahadevan, op.cit.
12. T.V. Mahalingam (1962), A new Brahmi inscription in the Tamil country, Silver Jubilee Volume, Archaeological Society of South India, 125-131.
13. T.V. Mahalingam (1974), Early South Indian Epigraphy, Madras.
14. R. Nagaswamy (1972), Kalvettiyal (Tamil), Madras, p.74.
15. I. Mahadevan (1966), Corpus of the Tamil-Brahmi Inscriptions, Seminar on Inscriptions, Madras, 57-73.

 Go to the top of the page

Home | Biodata | Biography | Photo Gallery | Publications | Tributes