Least-squares fitting methods are described, currently the most accura
te known, for determining the imaging aberrations and other important
parameters. The data required are either the image displacement or the
image diffractogram shape (or both), as a function of injected beam t
ilt; the analysis is largely common to the two cases. Explicit solutio
ns are given for some simple cases (e.g. coma-free alignment on the ba
sis of three images only) and some promising developments are consider
ed; the relevant aberration theory is set out in detail. Experimental
work reported in a companion paper has confirmed the presence of three
fold astigmatism at significant levels, but has not shown any evidence
of other higher order aberrations.