An obvious point - remember refinement exists to give you the best possible model so you need to look at the maps. And I guess that has to be a subjective assessment - if you SEE anything more clearly at the higher resolution - I would use that data, but if the maps do not improve discard it..
This is purely anecdotal, but I find there isnt much gained by using very incomplete higher resolution bits of data. However complete weak data can improve the scaling and temperature factor discrimination a good deal.. Eleanor On 21 May 2014 11:50, Kay Diederichs <kay.diederi...@uni-konstanz.de> wrote: > Hi Thomas, > > I too would use the data out to 2.55A, as you did. The main point is that > the 2.45A model produces a worse Rfree (30.16%) at 2.55A than the 2.55A > model does (30.03%). And this tendency is confirmed for the 2.35A model, so > there's no point in going beyond 2.55A. > > best, > > Kay >