H. Brass and G. Schmeisser have defined, for given lineEir functioncils R, new linear functionals fl" which Ccin often be treated in a simpler way than the original one. This was used by them to obtain error estimates for interpolatory quadrature formulae and deiiniteness criteria for linear functionals.The aim of this paper is a further development of the theory of this method, based on extensive use of divided differences and Peano kernels, and to show relations to other fields such as monotonicity of Compound quadrature formulae, error estimates for modified quadrature formulae for Cauchy principal value Integrals, error estimates for polynomial Interpolation and identities for B-splines. Especially, it is shown that the approach of Newman to show the monotonicity of Compound Newton-Cotes formulae is a special case of the reduction method, and that B-spline identities can be derived from the reduction method, by viewing B-splines as Peano kernels of divided differences.