Foundations of regular variation

The theory of regular variation is largely complete in one dimen- sion, but is developed under regularity or smoothness assumptions. For functions of a real variable, Lebesgue measurability su¢ ces, and so does having the property of Baire. We nd here that the preceding two properties have two kinds of common generalization, both of a combinatorial nature; one is exempli ed by �containment up to trans- lation of subsequences�, the other, drawn from descriptive set theory, requires non-emptiness of a Souslin 1 2 -set. All of our generalizations are equivalent to the uniform convergence property