Condensed Write‑Up

• Historical temperament cluster: Well/mean‑tone and related temperaments group tightly (e.g., werck_cl5 with schlick‑legros/schlick‑barbour).
• Equal‑division & parametric families: pipedum_*, egads, and neighbors inter‑link consistently across bin widths.
• Slendro‑like families: slendro*, sejati, and allied sets form a recognizable neighborhood.
• OT/JI heavy sets: Ratio‑rich scales cluster via interval‑content resemblance; mapping to EDOs will refine these further.
• Subset structure: Pentatonic‑within‑heptatonic relations appear frequently within an ±8¢ tolerance.

Distances are Jensen–Shannon over interval histograms (circular, mod period) with 50¢ bins for the overview; 25¢/12.5¢ runs confirm neighborhood stability.

Downloads

Histogram Previews

Neighborhood Diagrams

Reproduce Locally

Run these from the project root:

python scripts/scale_report.py --scl-dir assets/source/scl --out-dir reports --bin-width 50 --topk 8 --limit 200
python scripts/scale_graphics.py --scl-dir assets/source/scl --out-dir reports/svg --neighbors-json reports/neighbors.json --bin-width 50 --limit 30
python scripts/scale_subset_report.py --scl-dir assets/source/scl --out-dir reports --tol 8 --limit 200