The failure of the classical theories for swollen polymer networks was
clearly demonstrated by recent differential swelling experiments of G
ee, Eichinger and McKenna. We introduce a weaker form of the additivit
y assumption based on the separation of the gel into ''solid''-like an
d ''liquid''-like components. Using simple scaling arguments and the a
ssumption that the effective solvent quality for a crosslinked polymer
system is worse than for its uncrosslinked counterpart, we predict a
similar anomaly of the swelling activity parameter (dilation modulus)
in osmotic deswelling experiments in the semi-dilute range of concentr
ation. The relevance to Eichinger's and McKenna's experiments on conce
ntrated systems is discussed.