Dear Ethan, Just answering your last remarks,
very many years ago I've done tests (unpublished) where I tried to estimate the magnitudes Fbulk and PHASES Pbulk for the reflections due to bulk solvent. It came out that : - for high-resolution reflections the magnitudes are small (everybody knows) and phases are irrelevant to phases of F(atomic model) - for very low resolution reflections Fbulk are practically proportional to F(atomic model), and the phases are different by pi (as everybody knows), as it should be from the Babinet principle; so it works well here, say at resolution lower than 20 A (the tests were done with a middle-size protein experimental data) - however for reflections of the intermediate resolution say, 10-15A, Fbulk are already comparable in size with F(atomic model) but Pbulk are irrelevant to PHASES(atomic model); That means that at such intemediate resolution generating Fbulk, Pbulk by the Babinet principle is not a good idea. Obviously, my tests were not exhaustive, imperfect etc. However I think the results seem to be logical. Also working much later with Andrey Fokine we saw quite physical values for the parameters of a flat-mask model while for the Babinet-based bulk-solvent model Glykos & Kokkinidis (Acta Cryst D, 2000 ? I do not remember by hard) did not see and system. This also makes an indirect point for the flat-mask model. In that sens, the flat-mask model uses a more advanced information. Again, I think nobody claims that this model is perfect ; I hope a better model may be suggested with time. Best regards, Sacha Urzhumtsev