We quantify a limitation in the usual accounting of the finite-size
effects, where the leading [(Zα)4] and subleading [(Zα)5]
contributions to the Lamb shift are given by the mean-square radius
and the third Zemach moment of the charge distribution. In the
presence of any nonsmooth behavior of the nuclear form factor at
scales comparable to the inverse Bohr radius, the expansion of the
Lamb shift in the moments breaks down. This is relevant for some of
the explanations of the “proton size puzzle.” We find, for instance,
that the de Rújula toy model of the proton form factor does not
resolve the puzzle as claimed, despite the large value of the third
Zemach moment. Without relying on the radii expansion, we show how
tiny, milli-percent (pcm) changes in the proton electric form factor
at a MeV scale would be able to explain the puzzle. It shows that one
needs to know all the soft contributions to the proton electric form
factor to pcm accuracy for a precision extraction of the proton charge
radius from atomic Lamb shifts.