Perturbative evolution at small x

The conventional approach to perturbative evolution is illegal because the expansion in powers of $\alpha_s$ of the the DGLAP splitting matrix ${\bf P(z,\alpha_s)}$ breaks down at small $z$. The small-$x$ data for the proton structure function $F_2(x,Q^2)$ and its charm component $F^c_2(x,Q^2)$ show that a hard pomeron, with intercept close to 1.4, must be added to the familar soft pomeron. Conventional perturbative evolution may be applied to the hard-pomeron component and provides a striking test of perturbative QCD.