Please take a look at http://swift.cmbi.umcn.nl/gv/janne/index.html for an
updated version of the Chou and Fasman parameters. These are a bit better (also
expressed as proper, statistically underpinned preference parameters), but
the conclusions we will draw are still the same, and AMELK, VITWYF, PSDNG
remain valid donkey-bridges :-)