Given an exponential Lie group G, we show that the constructions of B. Currey, 1992, go through for a less restrictive choice of the Jordan-Hölder basis. Thus we obtain a stratification of g* into G-invariant algebraic subsets, and for each such subset Ω, an explicit cross-section ∑ ⊂ Ω for coadjoint orbits in Ω, so that each pair (Ω ∑) behaves predictably under the associated restriction maps on g*. The cross-section mapping σ : Ω → ∑ is explicitly shown to be real analytic. The associated Vergne polarizations are not necessarily real even in the nilpotent case, and vary rationally with ℓ ε Ω. For each Ω, algebras ε0(Ω) and ε1(Ω) of polarized and quantizable functions, respectively, are defined in a natural and intrinsic way. Now let 2d > 0 be the dimension of coadjoint orbits in Ω. An explicit algorithm is given for the construction of complex-valued real analytic functions {q1, q2, ⋯ , qd} and {p1, p2, ⋯ , pd} such that on each coadjoint orbit O in Ω, the canonical 2-form is given by ∑d pk ̂ dqk. The functions {q1, q 2, ⋯ , qd} belong to ε0(Ω), and the functions {p1, p2, ⋯ , pd} belong to ε1(Ω). The associated geometric polarization on each orbit O coincides with the complex Vergne polarization, and a global Darboux chart on O is obtained in a simple way from the coordinate functions (p1, ⋯ , pd, q1, ⋯ , q d) (restricted to O). Finally, the linear evaluation functions ℓ → ℓ(X) are shown to be quantizable as well.