Perturbative evolution: a different approach at small x

We propose an approach to DGLAP evolution at small x that circumvents the usual problem that a perturbation expansion is not valid there. The data for the charm structure function are important to motivate the method, and it describes them much more successfully than the conventional approach.