Can we understand the nature of the singularities that have to be admitted after a blow-up sequence that preserves the normal crossings locus of an algebraic (or complex-analytic) variety X? For example, every surface can be transformed by blowings-up preserving normal crossings to a surface with at most additional Whitney umbrella singularities. We will discuss general conjectures in arbitrary dimensions, and partial solutions. The techniques involve circulant matrices, elementary Galois theory and Newton-Puiseux expansion in several variables. We will discuss results in collaboration with Edward Bierstone and Ramon Ronzon Lavie.