The aim of this work package is to separate the combined test response into the separate internal and external length scale effect contributions. Material (length-enabled) performance = Combined test response adjusted for imposed test length scales It is known that, not only the material microstructures and test piece external dimensions, but also the length scale of the measurements contribute to the measured performance in a particular test. This is a large measurement challenge in real industry applications, since the component is often processed under special conditions at one length-scale in the factory, tested at another length-scale in the testing labs, and neither may be the same length scale as that experienced in service. Component performance = f {Material (length-enabled) performance, service length-scale, component size} The overall material performance will inevitably include the length scales from at least three sources: material microstructure sizes, test piece external dimensions and measurement (service) lengths. Quantifying the length scale contributions from each source will enable new designs of engineered components to maximise their mechanical performance by being tailored to a particular service condition. Another common contributor to mechanical performance is the residual stress introduced during component processing or service life. Residual stress has a direct influence on the perceived yield stress as the residual stress acts as a direct offset to any applied stress. The influence of residual stress upon indentation response is, however, disproportionately large and methods to normalise test response for residual stress influences need to be developed. This will enable the length-enabled mechanical response of the sample to be separated from the combined material + residual stress response of the test result. A side-benefit of this will be the ability to generate maps of relative stress state for a sample. The final stage of WP4 is to determine the feasibility of using an inverse application of the models from WP1 to infer a measurement value for otherwise difficult to measure properties from an input of test size effect and other more measurable length-scales in a sample.