Plomp&Levelt vs. Zwicker&Fastl?

[Marc Leman <Marc.Leman@xxxxxxxxx>]

Looking at the data on sensory dissonance (Plomp & Levelt, 1965, JASA) and
roughness (Zwicker & Fasl, 1990, Springer) we noticed that there is a
discrepancy in some of the values. For instance, P&L say that maximum
dissonance near 1000 Hz occurs at a 35 Hz difference (= 1/4 of a critical
bandwidth), while Z&F say it is 70 Hz (1/2 of a critical bandwidth). Other
values are similarly biased. One thing we noticed is that the probe tones
are different (dyads for P&L, amplitude modulated for Z&F). We don't see why
this should cause the difference. Have we overlooked something obvious or is
the data really inconsistent?

Marc Leman and Bill Sethares