There is well-known problem of significant systematic disagreements between data for reactions (γ, 1n), (γ, 2n), obtained at Livermore (USA) and Saclay (France) using the method of photoneutron multiplicity sorting. The averaged ratios RS/Lint of integrated cross sections obtained at Saclay and Livermore for 19 nuclei from 51V to 239U are equal to 0.84 for (γ, 2n) and 1.07 for (γ, 1n) reactions. For 75As RS/Lint ratios for both partial reactions are very close (1.22 and 1.21) but for 181Ta - are quite different (0.89 and 1.25). Using the objective physical data reliability criteria it was found that there are serious doubts in reliability of Saclay and Livermore data. The newly evaluated reliable cross sections disagree with experimental data. In addition to unreliable sorting of many neutrons between both partial reactions many neutrons were lost - on the case of 181Ta in 1n channel, in the case of 75As in both 1n and 2n channels.