Based on the experimental fact that the susceptibilities χ i (T) and the corresponding Knight shifts K i (T) (i=c,ab) are linearly related above certain temperature T χ * (>T c ), one normally draws a conclusion that a single Fermi component is operative. We show that this may not be generally valid. As a counter example we propose a two-component system were the susceptibilities are determined by a universal function f(T). The model consist of a Fermi component h + and a Bose component B ++ with triplet spin localized in CuO 5 sites, in chemical equilibrium with respect to reaction B ++ ⇌2h + , where f(T) gives fraction of bosons and 1−f(T) the fraction fermions. The susceptibilities above T * χ are given by adding the fermion and boson contributions in the form χ i (T)=χ i0 +A i [1−f(T)]+B i f(T), where χ i0 , A i and B i are T-independent. Clearly then χ c (T) and χ ab (T) are linearly dependent. If the bosons are localized within the CuO 6 octahedra or CuO 5 pyramids in the ab planes, rows of such tilted sites can explain the occurrence of stripes of localized charge and antiferromagnetic fluctuations in 2D CuO 2 planes.