We have analysed O-16 + O-16 elastic scattering at 350 MeV using a two
-step method (S(l) search followed by S(l)-to-V(r) inversion). We have
constrained the procedure so that only rather smooth potentials are d
etermined. The fits are nevertheless better than others involving dire
ct fits of parameterized V(r) to data. We present the results of two i
ndependent analyses which both yield V(r) closely resembling the type
A potential of Kondo et al. and also folding model V(r) for r greater
than or equal to 4 fm. The potentials differ in their range of sensiti
vity; for one potential changes in V(r) for r < 3.8 fm have no influen
ce whatever on the fit in strong contrast to certain folding model pot
entials. The general properties of the real components for both fits c
losely resemble folding model predictions. For both potentials, the im
aginary component that emerges is remarkably similar to that obtained
from a simplified Glauber model. We compare with results of previous t
wo-step fit and discuss strengths and dangers of two-step phenomenolog
y.