AbstractThe efficiency of a single fractionation step (i.e. of the partition of a given polymer sample between the coexisting phases which form when its solution demixes) is governed by the variation of the mass ratio of the individual P‐mers with chain length. These dependences have been measured for the system dimethoxymethane/diethylether/ polystyrene which exhibits lower critical solution temperatures by means of the Baker‐Williams method and by means of g.p.c. Their comparison with the results of previous investigation on systems showing upper critical solution temperatures reveals no thermodynamic differences in the aptitude of upper and lower critical solution temperatures for fractionation from dilute solutions.The extension of the molecular weight distribution into the low molecular weight region by the addition of oligomer, answered the question whether the monomer concentration in the coexisting phases is practically identical, as predicted by classical theories, or larger by a factor of nearly ten in the polymer‐rich phase, as extrapolated from the information concerning the polymer. The direct observation yields intermediate values which demonstrate that the variation of the logarithm of the partition coefficient with the degree of polymerisation P cannot be represented by a linear function with sufficient accuracy when a broader range of molecular weights is considered; in this case it is necessary to replace P by P2/3.