In previous work [T6], we employed the approach to Schubert polynomials by Fomin, Stanley, and Kirillov to obtain simple, uniform proofs that the double Schubert polynomials of Lascoux and Schützenberger and Ikeda, Mihalcea, and Naruse represent degeneracy loci for the classical groups in the sense of Fulton. Using this as our starting point, and purely combinatorial methods, we obtain a new proof of the general formulas of [T5], which represent the degeneracy loci coming from any isotropic partial flag variety. Along the way, we also find several new formulas and elucidate the connections between some earlier ones. BKTY]. This answer was extended in a type uniform way to the symplectic and orthogonal Lie groups in [T5]. The formulas of [BKTY, T5] rely in part on a theory of Schubert polynomials, which express the classes of the degeneracy loci in terms of the Chern roots of the vector bundles E • and F • . The desired combinatorial theory of Schubert polynomials, together with its connection to geometry, was established in the papers [LS, L, Fu1] (for Lie type A) and [BH, T2, T3, IMN1, T5] (for Lie types B, C, and D).In previous work [T6, §7.3], we employed Fomin, Stanley, and Kirillov's nil-Coxeter algebra approach to Schubert polynomials [FS, FK] to give simple, uniform proofs that the double Schubert polynomials of Lascoux and Schützenberger [LS, L] and Ikeda, Mihalcea, and Naruse [IMN1] represent degeneracy loci of vector bundles, in the above sense. Our main goal in this paper is to begin with the same definition of Schubert polynomials from [T5, T6] and, by purely combinatorial methods, derive the splitting formulas for these polynomials found in [T5, §3 and §6]. The latter results then imply the general degeneracy locus formulas of [T5]. The proof of the corresponding type A splitting formula from [BKTY] is essentially combinatorial; this is clarified in [T5, §1.4] and §1 of the present paper. As in [T5], our arguments depend on two key results from [BKT2, BKT3], which state that the single Schubert polynomials indexed by Grassmannian elements of the Weyl groups are represented by (single) theta and eta polynomials. The original proofs of these theorems used the classical Pieri rules from [BKT1], which were derived geometrically in op. cit. by intersecting Schubert cells. More recent proofs by Ikeda and Matsumura [IM], the author and Wilson [TW, T7], and Anderson and Fulton [AF2] use localization in equivariant cohomology (following [Ar, KK]) or employ other geometric arguments stemming from Kazarian's work [Ka]. However, the statements of the aforementioned theorems from [BKT2, BKT3] are entirely combinatorial, and it is natural to seek proofs of them within the same framework. The corresponding result in type A is the elementary fact that the symmetric Schubert polynomials are equal to Schur polynomials. The approach we take here begins by extending Billey and Haiman's formula [BH, Prop. 4.15] for the single Schubert polynomials indexed by the longest element in the Weyl group of G to the double Schuber...