This was an interesting and enlightening perspective. Some of my comments below:

1a. While Indian music is tonal, it is also strongly dependent on the intervallic distance between successive notes. As a consequence, a melodic sequence like
minor tone (10:9) * semitone (16:15)
yields a rather flat minor-third (32/27) which is missing in Table 1 because its percentage similarity value is 6.7%.

1b. A very common 'note' 45/32 does not appear in Table 1. While 45/32 may have a low percentage similarity value (5.3%) - it attains a strong position in the scale due to the melodic context imposed by its neighbours (invariably notes like a rather flat third or flat sixth); e.g. rag Miyan ki Tori.

1c. There is a whole host of 'sharp fourth' notes/intervals that are missing in Table 1: 25/18, 45/32 (discussed above), 64/45, and 36/25. These notes can have strong positions in a scale depending on melodic context.


2a. Some smaller intervals (like 25:24 which occurs between the 6/4 and 5/4 notes) are important in melodic music, even though their 'similarity' values are low (8% for 25:24). These do not show up in Table 1 either.

2b. In fact, choosing the 'top 60' intervals has caused quite a few intervals smaller than 100c to be dropped!


3. Merely using the percentage similarity as a metric suggests that the 27/20 note (8.5%) is more important than, say, 45/32 (5.2%). However, IMO, no Indian scale would use 27/20 as a position of rest as it would be construed as a mis-tuned fourth!


4a. Some very popular pentatonic rags don't end up in Table 2. For e.g. consider the simple but powerful rag Bairagi Bhairav (Southern Revati):
1/1 16/15 4/3 3/2 16/9 2/1
The similarity value comes out to be 38.26% well below the first 50 scales of Table 2.

4b. Or consider rag Kalavati, another popular and immensely pleasing rag:
1/1 5/4 3/2 5/3 16/9 2/1
whose similarity value is 35.46% - once again not in the top 50. Similar thoughts apply for other pentatonic rags employed in Indian music.

4c. It seems incongruous to me that a vast majority of the scales listed in Table 2 are not really encountered in any musical system, while quite a few that are very popular (at least in Indian music) fall well below the 40% similarity value.

4d. The only way to explain this discrepancy may be that in melodic systems the overall percentage similarity of the scale is not as important as the immediate intervals that are used in transitioning from one note to another.

4e. Given this, I'm tempted to posit that in Indian music, when moving from one note to the next separated by a pre-defined musical interval, preference is given to that specific intervallic ratio which has a higher percentage similarity metric (per Table 1) - though this preference may be overridden by the overall melodic context.


5. Finally, many South Indian Melakartas (generating scales) would automatically be disqualified under the percentage similarity criterion because often two or more semi-tonal intervals occur in sequence bringing down the overall percentage similarity of the scale. e.g. 6 Melakarthas each in the Indu Chakra, Rishi Chakra, Rutu Chakra, and Aditya Chakra = a total of 24 Melakartas, some of which are very popular even now (for instance, rag Nattai).


Thanks for your attention.
Roshan Kamath