The use of fibre spinning and related experiments is examined and the presentation of results in terms of a ＇＇spinning viscosity＇＇ is proposed. A re-examination of results for convected Maxwell and Jeffreys models and FENE dumbbell models shows the importance of the flow history prior to the spinning. Some undesirable ways of treating data are discussed. Out of this comes a discussion of the appropriateness of using viscosity or modulus as a measure of a material response and of using rate of strain or strain as a variable. The work is theoretical, but leads to a proposal for testing the ways in which experimental data are treated. A definition of a ＇＇spinning viscosity＇＇ is offered, and the idea that one should only obtain one value from one spinning experiment is emphasized.