Using the largest data set available, we determine the best values that the data at t = 0 (total cross sections and real parts of the hadronic amplitudes) give for the intercepts and couplings of the soft pomeron and of the ρ/ω and a/f trajectories. We show that these data cannot discriminate between a simple-pole fit and asymptotic log 2 s and log s fits, and hence are not sufficient to reveal the ultimate nature of the pomeron. However, we evaluate the existing evidence (factorization, universality, quark counting) favouring the simple-pole hypothesis. We also examine the range of validity in energy of the fits, and show that one cannot rely on such fits in the region √ s < 9 GeV. We also establish bounds on the odderon and the hard pomeron.