THE COUNT DOES NOT CHOOSE THE SOUND
Twenty-two counts.
Find another tuning.
The Sanskrit gives a count row and an instruction to lower Pa. Does that choose a single set of frequencies?
Keep the counts. Keep the octave. Change the size of a whole gap, then listen to what the numbers leave open.
Build a counterexample ↓Two choices.
The third follows.
Give a size in cents to each whole 4-count and 2-count gap. The 3-count size follows from closing the octave. No pitches are assigned to the individual śrutis inside a gap.
YOUR WHOLE-GAP SIZES
3-count gap calculated, not a third free control
A modern extra constraint. It fixes the 4-count size; the 2-count size can still move. These conditions do not stand for the text’s full consonance theory.
Shaded: 4-count > 3-count > 2-count gaps. Filled point: yours. Ring: saved witness. Dashed line: fourth/fifth lock. The shaded region belongs to this modern model.
The row 3–2–4–4–3–2–4 contains three 4-count gaps, two 3-count gaps and two 2-count gaps. Cents add; frequency ratios multiply. Displayed values are rounded.
Same counts. Different distances.
● Your tuning ┆ Saved witness
Each example plays Ma–Pa–Dha, then the same three with Pa changed. Synthetic sine tones; no ancient performance is reconstructed.
Inspect all note frequencies
Why can the fourth and fifth stay pure while another gap moves?
Write the whole-gap sizes as A (four counts), B (three), C (two). Closing the octave requires 3A + 2B + 2C = 1,200. In the initial row, Sa–Ma = A + B + C = 600 − A/2; Sa–Pa = 2A + B + C = 600 + A/2. Fixing these at 4/3 and 3/2 fixes A, but leaves C free while B compensates.
Pa’s change is A − B. It can vary even while those two initial consonances stay fixed. This is a family inside our chosen whole-gap model, not the complete space of ancient tunings. A successful candidate meets only the conditions shown here.
The counts agree.
Can you hear which is which?
Compare your two tunings without a label on the answer. A is the saved witness; B is your candidate. X repeats one of them, chosen afresh for each trial. Use a comfortable volume.
The setup button keeps the fourth/fifth lock, sets a ratio witness and moves the 2-count gap by ten cents. Or use your own pair above. Ma is an unchanged-note control when both tunings share the locked 4-count size.
Begin with a pair of tunings. Complete A, B and X at least once before answering; replay freely.
Your listening record
No revealed attempts yet. This record stays in this tab until you leave or clear it.
What does this exercise establish?
This is a self-guided A/B/X comparison using synthetic sine tones. The tunings, pitch, note sequence and level stay fixed inside a trial; X has exactly the sample data of A or B. The record counts only completed digital playbacks. It cannot establish that sound reached your ears.
Replays, immediate feedback and changing settings make this practice, not a controlled hearing study. Matches and misses do not establish a hearing threshold, ancient pitch sizes or a historically used tuning. Identical sample sequences are unscored. New trials and changed settings discard an unrevealed trial; revealed attempts remain, up to twelve per record. Answers are concealed in the interface, not protected against inspecting the program.
Read the Sanskrit instruction and inspect the whole printed page. The saved listening study includes the revealed record, exact tuning settings, playable A/B sounds, a reopenable JSON record and the complete source reading. Reopened playback counts and answers are labeled as recorded claims, not new listening in this tab. Opening a record replaces the current revealed record only after the whole file passes validation.
The instruction survives.
Our numbers are extra.
The count rows and lowering instruction come from the text. Cents, octave closure, equal size for repeated whole-count groups and the optional ratio lock are declared assumptions of this experiment.
What would this experiment fail to establish?
A candidate with ordered gaps and a lower Pa is a numerical witness for this count-row exercise. That does not establish a historically used tuning, an equal division of every śruti, or satisfaction of the text’s additional consonance and two-vīṇā conditions.
Grosset brackets the longer two-vīṇā passage on pp.28–29, and his note 36 on p.86 calls the interpretation approximate. This experiment leaves that problem open. The optional pure fourth/fifth lock supplies two modern numerical tests; it does not replace a study of all the ancient interval relations.
Read all eight excerpts and word notes · Grosset’s 1888 scanned volume · Source page identities
Keep your counterexample.
The reading records both witnesses, their checks and frequencies, embedded sound, Sanskrit and full printed evidence. Failed candidates keep their failed checks in saved work.
Saved links and JSON preserve both witnesses and the consonance lock.