The 2008 United Kingdom Research Assessment Exercise will involve the evaluation of thousands of individual research outputs. The Funding Councils set three criteria for those evaluations--Originality, rigour and significance--and required each output to be placed into a fivefold categorisation of excellence, using absolute rather than relative judgements. The various panels and sub-panels convened to undertake those evaluations were asked to provide their own interpretations of both the criteria and the categories. This paper analyses those, and finds that they are no less subjective than the generic descriptions provided by the central organisation.