Spiders use major ampullate silk (MAS) in web building and as a 'safety line' when escaping predators. In the former case, the fibre is spun at approximately 1 cm s(-1); in the latter case the rate is an order of magnitude faster. The different spinning rates do not result in identical microstructures, implying that the mechanical properties should not be identical either. However, neither conventional tensile testing, nor the characterisation of storage modulus with DMA yields reproducible data that could be used to correlate stiffness to the spinning rate. Any correlation is masked by the consequences of MAS having a cross-section that is non-circular, difficult to measure and highly Variable along the fibre length. Also, DSC shows that fibres contain bound moisture even after drying in a desiccator; fibre thickness affects the ability of samples to lose this moisture and gain stiffness on heating in the DMA. Thermal expansion data obtained with the DMA are reproducible below the MAS glass transition (approximately 160 degrees C) : these results do not depend on the accurate measurement of the sample cross section, they are insensitive to the moisture content of samples and they are not affected significantly by energy loss arising from friction between individual fibres in the specimens tested. The thermal expansion data suggest that the highest intrinsic stiffness of MAS is realised at spinning rates used in web construction. When the spider must elude a predator, it spins MAS that is more compliant.