The treatment of weak intensities in diffractometer data; the effect of truncating data

Abstract

Abstract It appears to be normal practice to exclude weak data from the least squares refinement, even when this data has already been collected on a diffractometer. What is more, there is no standard criterion for what is or is not a weak intensity. By means of a computing exercise based on two recently determined structures, it is pointed out that the effect of excluding weak data will improve the residual but this is cosmetic as it is accompanied by changes in coordinates, temperature factors, bond lengths and angles which are not only detectable but may exceed the e.s.d. in particular parameters thus leading to unquantifiable errors. In some cases changes in bond lengths and angles are apparently systematic with the degree of truncation. In addition truncating the data may lead to an increase in the computed e.s.d. in the parameters. This confirms previous work that data should not be truncated and all available data should be used for crystal structure determination.