Stress-free lattice parameter measurements in d0 cubes extracted from a nuclear power plant mock-up dissimilar metal weld have exhibited wide scatter among different orientations (upto 400 micro-strain). Since these variations directly affect residual stresses calculated in the weld, the Open University is investigating various factors that might contribute to the uncertainties in d0 measurements. The d0 sample must be large enough to completely envelop the gauge volume but small enough to be effectively stress-free. Often larger composite d0 samples are made by gluing together small coupons (more likely to be stress-free), but this raises uncertainties about the effect of glue, small gaps between coupons. The goal is to investigate the effect of d0 sample size and geometry relative to residual stress length-scale on d0 measurements and also to determine whether gluing is a reliable way.