Neoclassical tearing modes (NTMs) are plasma instabilities that can limit the performance of tokamaks and cause a termination of the plasma if allowed to grow. Systems to mitigate NTMs exist but have significant power requirements, which motivates further study of the mechanisms that lead to their growth in order to assist in the development of NTM avoidance strategies. NTMs typically require a seed magnetic island, above some threshold width, before they become unstable. The best available description of this threshold is the modified Rutherford equation (MRE) for NTM evolution; a combination of different models, which includes the effect of transport on NTM stability. Finite transport across magnetic field lines means that magnetic islands smaller than a critical width, w_c, do not completely flatten the pressure profiles and have a reduced bootstrap current perturbation, which leads to a threshold width, w_th.
This thesis describes novel measurements of NTMs with mode structure m/n=2/1 on the MAST spherical tokamak (ST), which have allowed a direct evaluation of the effect of transport on island behaviour for the first time on an ST. Temperature profiles obtained with the upgraded Thomson scattering system on MAST have been used to constrain the solutions of a heat transport equation for a magnetic island, allowing the experimental determination of w_c, an important parameter in the MRE. The measured value of w_c = 0.7\pm 0.2cm obtained for an ensemble of MAST discharges is used in an analysis of the MRE for 2/1 NTM onset and saturation on MAST. By using a probabilistic method for parameter and error estimation, which takes account of the experimental uncertainty on measured equilibrium parameters, it is found that the temporal evolution of the island size is well described by marginally, classically unstable NTMs (that is, Delta'>0) with strongly destabilising bootstrap current and stabilising curvature terms. Finally, an analysis of two beta ramp-down discharges is presented, in which the measured w_c value explains the observed threshold width well.
