When outliers are different

When does the presence of an outlier in some measured property indicate that outlying object differs qualitatively, rather than quantitatively, from other members its apparent class? Historical examples include many types supernov\ae\ and short {\it vs.\/} long Gamma Ray Bursts. There may be only one parameter outlier, so principal component analyses are inapplicable. A qualitative difference implies has a characteristic scale, hence distribution cannot power law (that can have no such scale). If is objects differ quantitatively. The applicability to empirical tested by comparing most extreme member next-most extreme. probability their ratio calculated, compared data for stars, radio X-ray sources, fluxes, fluences rotation measures Fast Radio It found with high statistical significance giant outburst soft gamma repeater SGR 1806-20 differed qualitatively lesser outbursts FRB 200428 (by location Galaxy), but supernova remnant models measure 121102 not, statistically significantly, outlier.