Abstract Levine has argued that Dummett’s distinction of analysis and decomposition was mistakenly attributed to Frege, but could be correctly attributed to Russell. Levine’s argument presupposes that, in one central respect, propositional functions had for Russell the theoretical significance that incomplete expressions had for Frege. I will argue that this was not the case, and that this fact points to a fundamental contrast between the two philosophers’ conceptions of analysis.