Glass Condensate and colour sources ∗

In the recent years we have seen a lot of activity around systems and experiments, like DIS at HERA or the heavy-ion experiments at RHIC, involving a large number of partons due to the high enery and/or the high number of participants of those experiments. The main problem in this regime is that of the high parton densities. In fact, in most of the models of multiparticle production, two contributions to the multiplicity are considered: one proportional to the number of participant nucleons, Npart, and a second one proportional to the number of collisions, N 4/3 part. In order to get the right multiplicities at RHIC it is necessary to lower the second contribution. A possible mechanism for this is the saturation. Here, I am going to review the saturation of parton densities in the initial state, in two different frameworks: the Colour Glass Condensate and the string clustering.