A Rock Wall at Kudumiyanmalai in Tamil Nadu Carries a Musical Score of Seven Sections and About 2,400 Notes, Read by Scholars Since 1904 and Dated to the Seventh or Eighth Century

On a hillside at Kudumiyanmalai in Pudukkottai district, Tamil Nadu, a score of seven sections of notes is cut into the rock in Sanskrit, in Grantha script, on a wall about 13 feet by 14 feet. A University of Cambridge music faculty abstract from 2024 counts some 2,400 notes in it and calls it the earliest surviving example of musical notation in South Asia. Epigraphists have been reading the wall since 1904, and when it was cut, and for which king, is still argued.

What is cut into the wall

The inscription lies to the south of a rock-cut shrine called Melakkoyil, beside a carved image of Ganesha, on the slope behind the Sikhanathaswami temple. P. R. Bhandarkar edited it in volume 12 of Epigraphia Indica (1913 to 1914) from an inked paper impression supplied by the epigraphist H. Krishna Sastri, and writes that it was discovered in 1904. Puratattva and Swarajya both give the wall as 13 by 14 feet.

The text opens with a salutation to Shiva. Seven sections follow, each headed by the name of a mode: Madhyama-grama, Shadja-grama, Shadava, Sadharita, Pancama, Kaishika-madhyama and Kaishika. The DHARMA project, a digital edition of inscriptions funded by the European Research Council, transcribes the same seven headings. Bhandarkar read each heading as saying that the notes beneath it were “produced by four strikings” of a string.

Each section is a block of four-note groups, arranged in sub-sections of sixteen, with one sub-section to a line, Bhandarkar writes. Every group in a sub-section ends on the same note. The sub-section that ends on the mode’s home note, the nyasa (final resting note), comes last, and the others are ordered by where their ending note falls in the scale.

The notes are written as syllables: the first letter of a note’s name plus one of four vowels, so that “sa” appears as sa, si, su or se. Bhandarkar could not say what the vowels meant and guessed they marked different ways of plucking. The right-hand ends of the sixth and seventh sections lie under the base of a stone hall built in front of the rock, which hides only the last few notes of each line.

The music, and what it was for

Bhandarkar compared the score with the treatises. He found that its notes fit each mode as the old texts define it, with differences between the texts themselves, and that it agreed better with the Naradi-siksha, a treatise of uncertain date, than with the Sangita-ratnakara of Sarngadeva, which he places about six centuries after the inscription. He also found that none of the seven names is treated as a raga in the chapters on music in Bharata’s Natyashastra, and he took the wall to be their earliest record. (The Natyashastra has its own tuput article.)

Richard Widdess, of SOAS University of London, returned to the score in an article of 36 pages in Musica Asiatica, volume 2. The MusicResearchLibrary catalogue calls it a rigorous study with a convincing theory. Pages of the article visible online show him testing whether the vowel endings mark rising and falling intervals between notes, by counting how often particular note pairs occur with particular vowel sequences. The Cambridge abstract describes the inscription as explicitly didactic, probably exercises in modal composition and instrumental playing.

Not everyone reads the sections as ragas. T. K. Venkatasubramanian, in a Sahapedia interview, holds that the wall records gramas and jatis (older frameworks of scales and melodic families) rather than developed ragas, and that it has no ascending or descending patterns.

Eight and seven

A short Tamil line follows the Sanskrit. The DHARMA edition translates it as “These are appropriate for eight and for seven.” Widdess takes the numbers to concern rhythm: Puratattva reports him as reading them as beats, and Swarajya as strikings. Other readers tie the numbers to the string count of the instrument the notes were written for.

A label beside the Ganesha image reads Parivadini. The South Indian Inscriptions volume glosses it as a lute with seven strings and takes it as evidence that the notation was meant for the vina. Sahapedia’s Kudumiyanmalai walkthrough, with J. Raja Mohamad, says inscriptions at Kudumiyanmalai, Malayakkoyil and Thirumayam mention the parivadini vina, which he says had seven or eight strings.

The same volume records a second, badly worn inscription of the kind on the walls of the rock-cut Shiva temple at Tirumayyam, where the names of notes can still be read, again in seven sections.

Seventh century or eighth

No date is written on the wall, so dating rests on the shapes of the letters. Bhandarkar assigned the characters to about the 7th century AD. South Indian Inscriptions, volume 12, also gives the 7th century. Jouveau-Dubreuil, in volume 1 of Pallava Antiquities (1916), puts it probably at the beginning of the 7th century, in characters similar to those at Trichinopoly. The Cambridge abstract of 2024 calls it 7th-century, and Puratattva adds that K. R. Srinivasan accepted Bhandarkar’s date.

A later group argues for the eighth. Puratattva lists K. R. Venkata Rama and K. V. Soundara Rajan for the 8th century, and D. Dayalan for the second half of that century, with the rock-cut cave in the first half. Swarajya gives circa 8th century. Raja Mohamad, in the Sahapedia walkthrough, places it in the 8th or early 9th. The earliest dated record in the cave, Puratattva says, is an inscription of Varaguna-varman I from 787 to 788 AD, which is not a foundation record, so the cave itself is older than that.

Which king

The Sanskrit closing lines say the treatise on notes was made for the benefit of pupils by a king who was a pupil of Rudracharya and a devotee of Shiva. In the DHARMA edition the letters of the word for king are in parentheses, which in an edition marks letters the editors supply. No name is given.

South Indian Inscriptions, volume 12, files the record under the Pallava king Mahendravarman I. It cites the 7th-century letters and a title, Sankirnajati, that it says Mahendravarman bore, and adds that he had distinguished himself in architecture, literature and drama. Puratattva and Sahapedia credit Jouveau-Dubreuil with first proposing Mahendravarman, and Sahapedia says T. N. Ramachandran backed the idea in a 1931 article, “Mahendravarman, the Royal Artist”. The Dubreuil volume I opened is more careful: it says nothing shows that Kudumiyamalai was under Mahendravarman’s direct rule. I did not find where he made the authorship suggestion. If the attribution holds, the author belonged to the same Pallava line that cut the shore temples at Mahabalipuram and built the Kailasanathar temple at Kanchipuram.

The objections, as Swarajya sets them out, are geographical and architectural. The Kaveri was the southern limit of Pallava rule in Mahendravarman’s time; no other Pallava inscription is found in the cave; the square base of the lingam is a Pandya feature; and Pallava caves usually carry royal praise-verses, which this one lacks. Swarajya proposes the Pandya king Varaguna instead, and Puratattva reports that Soundara Rajan suggested a king named Gunasena might be the Pandya Varagunavarman I. The Tamil Nadu Temples Project reads parama-maheshvara in the closing lines as a king’s name, Parama Maheshvaran, and says the earlier Mahendravarman attribution is now disproved. Sylvain Brocquet’s French translation, printed on the DHARMA page, renders the words as a description, “sectateur passionne de Mahesvara,” a devotee of Shiva. Who Rudracharya was is also open: Puratattva records that T. N. Ramachandran and J. R. Marr identified him with Rudrata, the author of the Sringaratilaka.

Where the work stands

The DHARMA edition lists at least five editions of the text: Bhandarkar’s of 1913 to 1914; the Pudukkottai Darbar’s of 1929; T. V. Mahalingam’s Inscriptions of the Pallavas (1988); Brocquet’s French translation (1997); and the Pallava Art volume by Lockwood and colleagues (2001). The DHARMA page shows its own digital version as last modified on 29 September 2026.

On 3 December 2024, Richard Widdess, Marcus Pearce of Queen Mary University of London, and Alberto Alcala Alvarez and Pablo Padilla of the Universidad Nacional Autonoma de Mexico presented a Cambridge seminar on a project to build a digitised text of the inscription, to test Widdess’s earlier reading with statistics, and to ask whether the notation could be a fingering system for harp.