Perhaps one of the most puzzling areas in syntax-semantics research concerns how sentences with comparative "more" are constructed and interpreted. I present a number of novel observations and generalizations, with an eye to understanding the roles of the lexicon and syntax in constraining the distribution of so-called "monotonic" versus "non-monotonic" scales. I consider how existing theories might be extended to (minimally) describe the new data, and propose an alternative aimed at better explaining it.