This thesis reports work in three areas of X-ray crystallography. An initial chapter describes the structure of a protein, the methods based on the use of X-rays and computer analysis of diffraction patterns to determine crystal structure, and the subsequent derivation of the structure of part or all of a protein molecule. Work to determine the structure of the protein cytokine receptor-like factor 3 (CRLF3) leading to the successful generation of a structural model of a significant part of this molecule is then described in Chapter 2. A variety of techniques had to be deployed to complete this work, and the steps undertaken are described. Analysis was performed principally using phaser, using maximum likelihood methods. Areas for improvement in generating non-crystallographic symmetry (NCS) operators in existing programmes were identified and new and modified algorithms implemented and tested. Searches based on improved single sphere algorithms, and a new two-sphere approach, are reported. These methods showed improvements in many cases and are available for future use. In Chapter 4, work on determining the relative importance of low resolution and high intensity data in molecular replacement solutions is described. This work has shown that high intensity data are more important than the low resolution data, dispelling a common perception and helping in experimental design.