On the analytical approximation to the GLAP evolution at small x and moderate Q²

Comparing the numerically evaluated solution to the leading order GLAP equations with its analytical small-x approximation we have found that in the domain covered by a large fraction of the HERA data the analytic approximation has to be augmented by the formally non-leading term which has been usually neglected. The corrected formula fits the data much better and provides a natural explanation of some of the deviations from the $\sigma$ scaling observed in the HERA kinematical range.