Change the first note.
Then change one string.
Early Indian music theory distinguishes a new order of notes from a change in their tuning. Try both, using the two grāmas in the Nāṭyaśāstra.
Seven notes. Twenty-two counted intervals. Lower just Pa and two adjacent groups exchange a count.
Keep the notes. Change the order.
Choose where the ascent begins. Sa keeps its pitch; notes that cross the octave double theirs.
Counts, not pitch distances. Each cell is one counted śruti; equal widths do not imply equal sounds. Numbers sit in the gaps they count. Six named notes stay fixed. Pa moves in plum.
Sine tones under the pitch assumption below. No sound plays until you ask.
A count is not yet a frequency.
The count row alone leaves the tuning undetermined. Compare two explicit modern assumptions; neither is a reconstruction of the whole ancient theory.
Pitch distance in cents: the upper row is ṣaḍja-grāma, the lower madhyama-grāma. An octave is 1,200 cents. Only Pa changes.
Find another tuning → Keep the count rows and octave; build your own pitch example. Can the initial fourth and fifth stay pure while Pa’s lowering changes?
Can you hear which tuning? → Compare a concealed repeat with two frozen examples, then inspect the pitches behind your answer.
What do the two pitch assumptions mean?
The ratio example assigns 9/8 to a whole four-count gap, 10/9 to three, and 16/15 to two. It does not assign pitches to every internal count cell. Their seven gaps multiply to exactly 2, an octave. It is a calculation chosen for this lesson, not a ratio table quoted from Bharata.
The equal-division counterexample instead assigns 1,200/22 cents to every count. It preserves this count row while giving different pitches. It shows why the number 22 alone does not establish equal temperament. It is not offered as an ancient tuning or as satisfying the text’s additional consonance requirements.
The longer two-vīṇā experiment has textual difficulties. We keep to the named count rows and the single Pa operation; we do not claim either sound model resolves that experiment. Inspect the evidence.
Read the operation.
Sanskrit, a Roman reading (IAST), and new English. Begin with the instruction to lower Pa, then follow the counts in the two arrangements.
Edition, transcription and uncertainty
We read chapter 28 of the Nāṭyaśāstra, attributed to Bharata, through Joanny Grosset’s 1888 Sanskrit edition and French translation. The work is composite and transmitted in differing manuscripts; this lesson assigns no exact date to the passage.
Grosset’s Sanskrit is normalized to IAST; the Devanagari here renders that chosen reading. Full printed pages retain his apparatus. The count verses also appear in Śivadatta and Parab’s 1894 edition at different verse numbers. The final half-verse definition of mūrchanā comes from that edition.
Grosset p.21 says brackets mark doubtful passages. He brackets the longer two-vīṇā demonstration on pp.28–29; note 36 on p.86 calls his interpretation approximate. Our brief instruction and count rows remain distinguishable from that unresolved demonstration.
1888 scanned volume · 1894 scanned edition · Source and model notes · Page identities
The octave closes.
The wrong note moves.
The named-note verses assign Sa four counts, Re three, Ga two… Does each count belong to the preceding interval or the following one? Try the wrong attachment. Check it against the instruction śrutyapakṛṣṭaḥ pañcamaḥ kāryaḥ: Pa is to be made lower by a śruti.
Ma–Pa–Dha before and after the count change. Synthetic sine tones, not an ancient performance.
Inspect every note frequency
The main bench above always uses the preceding-interval reading. This diagnostic keeps Sa fixed and uses the same modern pitch assumption and Sa setting as that bench. Equal-width cells count groups, not equal pitch steps.
Save the comparison JSON, copy its link, or save the illustrated reading below to keep this placement choice.
Keep the comparison.
Save the chosen starting note, tuning and pitch assumption. The illustrated reading includes all passages, source pages, fonts and a listening example in one file.
The saved reading works as a local file. The WAV is synthetic sine audio of Ma–Pa–Dha before and after lowering Pa.