Headline finding
Per-tune statistics
mean |leap| = mean absolute interval (semitones); % β₯ P4 = share of intervals a perfect 4th or wider (β₯ 5 st); % β€ M3 = share that are 3rds or smaller steps (1β4 st); osc = chained direction-reversing wide leaps. Remainder to 100% = unisons (repeated notes).
Significance tests
Aggregate interval distribution (ιε vs ζ±ε)
Each bar = that interval's share of the region's intervals. The crossover is the whole story: ζ±ε peaks on 3rds (m3/M3), ιε peaks on 4ths/5ths (P4/P5).
Wide-leap share by tune
Mean leap vs wide-leap %
Each point is one tune. The clusters separate on the vertical axis (how often wide leaps occur), far more than on the horizontal (average leap size).
Oscillating leaps β the ιε saw-tooth
A leap up answered by a leap down (both β₯ P4). This chained, direction-reversing motion is the ιε signature and is almost absent in ζ±ε.
Interval lines on the 12-tone clock β 4 vs 4, pooled
Every consecutive interval drawn as a chord on the 12-tone clock (tonic C at 12 o'clock; all tunes normalized to 1 = C). Line darkness is normalized by interval count, so the two wheels compare fairly: ιε rides the orange 4th/5th crossings, ζ±ε the blue-green stepwise rim. Interactive play-&-trace versions live on the landing page and each melody page.
3-mer triangles β 4 vs 4, pooled
Every run of three consecutive notes drawn as a triangle between its pitch classes, colored by its widest interval; oscillating-leap triplets (both β₯ P4, direction reversed) are outlined boldly β 21 in ιε vs 2 in ζ±ε, matching the oscillation counts above.
3-mer chord histogram β which triangles each region rides
Each 3-mer reduced to the chord it outlines (1 = C; β = a 3-mer confined to two notes, e.g. DβG covers DβGβD and DβDβG). Note the near-disjoint vocabularies: ιε on the suspended/quartal Csus2, DβG, DβA, Gsus2; ζ±ε on the tertian Am7(no5), Em7(no5) and stepwise GβA. Unlike the interval distribution above, the unit here is a whole three-note window β it measures what the melody camps on, not single steps.