We review the evidence for the presence of a hard singularity in soft forward amplitudes, and give an estimate of its trajectory and couplings.In 1998, Donnachie and Landshoff showed that DIS data at small x can be described by the superposition of two Regge exchanges : the soft and the hard pomeron, and that both are well approximated by simple poles [1]. Their fit to HERA data then leads to the conclusion that the soft pomeron is a higher-twist contribution. It was further suggested [2] that the DGLAP evolution has to be performed in a non-trivial manner : as it is of perturbative origin, the singularity that it introduces at J = 1 should be considered only if one is far away from it, as otherwise perturbation theory breaks down. Applying this philosophy to the evolution leads to a prediction of the Q 2 dependence of the hard pomeron from DGLAP evolution, and to a successful description of F 2 and F L [3,4], provided that the gluon distribution is associated with the hard pomeron contribution. The obtained fit can then be extended to higher x [3]. This leads to the conclusion that the data at large Q 2 and t = 0 contain a sizable contribution from a hard pomeron with intercept α h (t = 0) ≈ 1.4.Furthermore, the same idea can be used to describe F c 2 [5] as well as vector-meson photoproduction [6]. In this case, the main characteristic of the data is a modification of the energy dependence of the cross section as the vector-meson mass increases, and a flattening of the t dependence at high t. Both effects can be interpreted as a signature of the contribution of a hard pomeron, provided it has a rather flat trajectory:The question remained however to understand why such a hard pomeron had not been seen before, and what happened to it at t = Q 2 = 0. The first remark is that it is