Dispositions depend on "categorical" facts definitionally and pedagogically. Must they always depend on them also ontologically for "grounding"? Does there really have to be an ultimate "bottom level" of matter, and must it be "categorical"? The concepts microphysics supplies, however, are dispositional in meaning. What predicates aren't? Besides "shaping" and "locating" predicates, predicates expressing degrees of similarity and dissimilarity are nondispositional enough in meaning: but the predication of all these features of things depends upon other features for these to bound (...) and to relate to one another comparatively. Faced with the uncomfortable alternative of "dispositions all the way down" Simon Blackburn proposes antirealism. Possibly, though, predicates' dispositionality or categoricality can be relative to a given level of the organization of matter. (shrink)