Abstract

The accuracy of the determination of X-ray intensities, and hence structure factor F-values, is of crucial importance to studies of the solid state, particularly at the present time when computational facilities allow exhaustive analysis of experimental data against theoretical models. Experimental errors are far less easy to estimate than appears from consideration of published individual experiments at a first, or even more careful, consideration; methods of both estimating and correcting for them need very careful elucidation. By far the best way to do this - and probably the only way that will yield reliable information as to where we are in this ill-defined field - is to organize group projects specifically to allow estimation of the magnitude of overall error, the identification of individual sources of error where possible and hence, the detection of specific experimental features, which should be either carefully assessed in each experiment to allow the magnitudes of errors to be kept as low as possible, or actually physically corrected. It is helpful in this regard to allow error-sources to be thought of as stabilized or variable according to the type of project planned. It is obvious that careful planning of a series of projects with different characteristics will be an evan more powerful tool for the investigation of error-sources in depth. In this paper, two projects are considered in some detail - one originated by the IUCr the other by the ACA - both having considerable similarities but also instructive differences. A method of comparing and also of editing the projects is considered using the correlation R factor |Rij| where |Rij| = |Fi - Fj| /1/2 |Fi + Fj|. This is seen to allow simple but important deductions.