What can we learn by comparing experimental and theoretical-computational X-ray scattering data?

In this letter we use X-ray scattering data of liquid water, as obtained by different experimental and theoretical-computational procedures, to address the problem of quantitative modeling of the scattering signal in liquids. In particular we investigate the accuracy of well optimized water models in reproducing top level X-ray experimental results and compare experimental data variations with the ones given by different theoretical-computational models. Results show that the experimental scattering data have an intrinsic noise which is comparable to the deviations of the theoretical-computational signals, hence suggesting that no reliable refinement based on scattering data is possible for such models.