What are the specificities of the thermomechanical behaviour of tridimensional polymers in glassy state or in the glass transition region? Despite the fact that it is often claimed that ＇linear polymers are softer and tougher than tridimensional ones＇, detailed studies, for instance of elastic properties (1) indicate that these specificities, if they exist, are very scarce. A possible cause of systematic difference between both polymer families could be the crosslinking effect on relaxation spectrum. As a matter of fact, chain motions must be perturbed in immediate vicinity of crosslinks. Thus, if the width of the relaxation spectrum expresses the diversity of the situation of chain segments, it is expected to be higher in tridimensional than in linear polymers. The best way to obtain information on the relaxation spectra is dynamical testing and study of complex modules E* in the frequency domain from Cole-Cole plots. It seemed interesting to use this approach on a family of networks based on the same epoxide-amine pair : diglycidyl ether of bisphenol A(DGEBA)-diaminodiphenyl methane (DDM) with various amine/epoxide molar ratios. In such a family, it is possible to obtain large variations of the crosslink density with an almost constant chain stiffness. In non-stoichiometric samples, there is a great diversity of chain segment situations since certain ones are directly connected to the crosslinks whereas others are relatively far away.